No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
The accuracy and precision of underwater measurements of length and maximum body depth of Southern Bluefin Tuna (Thunnus maccoyii) with a stereo-video camera system
Abstract
The accuracy and precision of in situ stereo–video measurements of the snout to fork length (SNFL) (range 830–1412 mm) and maximum body depth (MBD) (range 228–365 mm) of free-swimming southern bluefin tuna (SBT) (Thunnus maccoyii) were tested by filming live fish in sea cages immediately prior to harvest. Stereo–video measurements of the SNFL of 54 fish produced an average error of 1.72 mm (relative error of 0.16%), while an average error of 1.37 mm (relative error of 0.51%) was recorded for measurements of MBD from 47 fish.
A procedure was developed to maximise the accuracy and precision of measurements of the SNFL and MBD from a single SBT over sequential images to avoid the underestimation of SNFL and overestimation of MBD due to sinusoidal changes in body form associated with fast swimming.
The results demonstrate the potential of stereo–video systems to non-destructively make counts and measurements of tuna and other fish in both wild fisheries and mariculture situations, without the need to capture and handle them.
... The microcontroller unit orchestrates seamless communication, processing, and analysis of the acquired data, enhancing sand pot management. By doing so, it not only optimizes sandblasting operations but also empowers organizations with the means to proactively manage maintenance and reduce operational costs [3]. Figure 1 provides an in-depth representation of the streamlined architecture used to measure sand levels, leveraging advanced sensor technologies as a robust method for real-time sand level monitoring. ...
... The system's effectiveness depends on data collection from sand level sensors, a series of sophisticated devices meticulously calibrated to provide accurate and precise readings [3] of sand levels in the sand storage unit. These advanced sensors are capable of collecting real-time data, providing a foundation for further operations. ...
... The system prevents overfilling and depletion of sand pots, leading to optimized resource utilization. Of which, resources are an integral part of industries as it contributes to the performance of the select industry [3]. Without excellent management of the sand as resources of the foundries, problems might occur internally for the foundries such as profit loss, system malfunction and damage to the sand pot. ...
... Stereo-video photogrammetry has been used for underwater measurements since the 1980s (see the review by Shortis, Harvey, and Seager 2007) for many purposes such as estimating biodiversity, abundance, fish behavior, reactions to anthropogenic pressures (e.g., Griffin et al. 2016) and along with other techniques like paired lasers, has been utilized to obtain in situ length estimates for marine megafauna (Klimley and Brown 1983;Rohner et al. 2015;Perry et al. 2018). Although it has a long history of use in aquaculture environments (Harvey et al. 2003;Silva, Aires, and Rodrigueset 2023;Naiberg et al. 1993), this technique is scarcely reported for use in aquaria, being mentioned by Harvey et al. (2003) in a study where 17 reef fish from six species were measured in an aquarium and posteriorly in situ, by stereo-video and divers allowing the comparison of these two methods. Later, Letessier et al. (2013) reported this technique in a study where Antarctic krill was measured, and more recently, Ikeya et al. (2022) published a study reporting the measurement of six individuals of catfish using stereo-video and the direct linear transformation method. ...
... Stereo-video photogrammetry has been used for underwater measurements since the 1980s (see the review by Shortis, Harvey, and Seager 2007) for many purposes such as estimating biodiversity, abundance, fish behavior, reactions to anthropogenic pressures (e.g., Griffin et al. 2016) and along with other techniques like paired lasers, has been utilized to obtain in situ length estimates for marine megafauna (Klimley and Brown 1983;Rohner et al. 2015;Perry et al. 2018). Although it has a long history of use in aquaculture environments (Harvey et al. 2003;Silva, Aires, and Rodrigueset 2023;Naiberg et al. 1993), this technique is scarcely reported for use in aquaria, being mentioned by Harvey et al. (2003) in a study where 17 reef fish from six species were measured in an aquarium and posteriorly in situ, by stereo-video and divers allowing the comparison of these two methods. Later, Letessier et al. (2013) reported this technique in a study where Antarctic krill was measured, and more recently, Ikeya et al. (2022) published a study reporting the measurement of six individuals of catfish using stereo-video and the direct linear transformation method. ...
Accurate collection of biometric data is important for understanding the biology and conservation of marine organisms, including elasmobranch and teleost fish, both in nature and controlled environments where monitoring marine specimens' health is mandatory. Traditional methods involving specimen capture and handling are invasive, stressful, and disruptive. Some techniques like underwater visual census or laser photogrammetry have been used for noninvasive data collection, but they have limitations and biases. The application of stereo‐video photogrammetry through the use of diver‐operated stereo‐video systems (stereo‐DOV) is a noninvasive method that overcomes these challenges, providing highly accurate measurements. It has become popular for species monitoring, studying anthropogenic impacts, and assessing length distributions. However, this technique is still uncommon and barely reported in aquarium settings. This study describes an innovative pilot study targeting multiple species carried out in a Public Aquarium, using a low‐cost house‐made device. The results revealed that measuring more than 100 individuals in approximately 1 day's work is possible. Total and fork lengths were estimated using specific software for 31 teleost and 16 elasmobranch species and compared with real measurements for the available species. Despite technical limitations that must be reviewed for application in future studies that resulted in high root mean square (RMS) values (> 20 mm), differences between methodological approaches revealed a minimal discrepancy (1.37%–5% in large sharks and rays and 1.8%–5.5% in teleost fish). This technique has time and cost requirements, but might represent a major advance in husbandry and in the contribution to conservation that ex situ studies can provide.
... Effective management of aquaculture farms requires precise biomass and fish length estimations during rearing (Harvey et al., 2003;Voskakis et al., 2021). Fish biomass is derived from the total number of fish counted in a specific volume of water multiplied by the average weight of the fish sampled (Harvey et al., 2003). ...
... Effective management of aquaculture farms requires precise biomass and fish length estimations during rearing (Harvey et al., 2003;Voskakis et al., 2021). Fish biomass is derived from the total number of fish counted in a specific volume of water multiplied by the average weight of the fish sampled (Harvey et al., 2003). Measuring biomass can help predict daily intake demands to avoid under or overfeeding (Lopes et al., 2017). ...
Estimating fish growth in real time has many benefits for indoor aquaculture farms, such as saving labor time and costs, reducing water pollution during feeding, improving feeding activity and determining when to harvest. Hence, this study proposed a visual‐information‐based method for measuring olive flounder (Paralichthys olivaceus) length, using accurate growth tracking for efficient aquaculture management. Using two cameras, an light‐emitting‐diode (LED) grid was placed at the bottom of the water tank to measure fish length. The pixels unit from the fish length in the captured image was converted to centimeters based on the relationship of a pre‐built dataset. A total of 180 lengths were calculated using images captured by the cameras. The average length of each fish acquired from the cameras was calculated separately, and Lagrange's interpolating polynomial algorithm was implemented to calculate the overall length of each fish. This method reduced the computational complexity, and results were obtained more rapidly and in a user‐friendly environment. The power model generated the length–weight relationship, which allowed us to estimate the body weight of each olive flounder based on the length. The proposed approach enabled us to calculate the length of olive flounders with a highly accurate R² of 0.995.
... Fish biomass is usually estimated from the total number of fish counted in a pond or cage multiplied by the average weight of a fish sample [12]. Capture-dependent methods are occasionally used to know the length and weight of the farmed fish, but the inaccuracy of the results, along with the stress and the potential increase in fish mortality, has led researchers to develop new non-intrusive estimation methods [5,[13][14][15]. ...
... The second approach involves computer vision, which is rapidly developing as a system for obtaining fish measurements without any kind of manipulation, using stereo vision underwater cameras. This technology uses images to measure the fish size in aquaculture facilities through computer image analysis [3,12,[25][26][27][28][29]. These two techniques have proven to be a good adjustment in fish size measures, but both usually need to use a previously calculated length-weight relationship to provide weight estimates from TS or stereoscopic images [11,30]. ...
Different non-intrusive methods have been developed to estimate fish biomass, which is a determinant factor for aquaculture farming management. Length–weight conversion is a crucial parameter for accurate biomass estimation. However, the potential environmental and seasonal variations in fish length–weight relationships are rarely considered. In this study, we examined seasonal variation in length–weight relationships for European sea bass (Dicentrarchus labrax) from two farming systems subject to different salinity and temperature conditions: inland ponds and offshore cages. The results showed significant differences in intercept and slope between the two types of facilities studied, as well as between the same seasons for both facilities. This highlights the need to use specific length–weight equations to obtain accurate biomass estimation based on fish length data.
... Until recently, none of these methods could specifically link behavioral observations to body size and interactions with the environment. However, stereo-video cameras (SVC) are a video-based in-water non-invasive approach that has vastly grown in acceptance as a cost-effective method for estimating animal body size [22][23][24][25][26][27]. SVC systems comprise two cameras that record overlapping perspectives to create a three-dimensional image allowing for length measurements to be estimated. ...
... One potential advantage to using the SVC to discern marine animal behavior is the ability to observe individuals in their natural, relatively undisturbed habitat, as well as potentially assess an animal's response to anthropogenic disturbances. Fortunately, advances in stereovideo photogrammetry make measuring in-water body length easier and more accurate compared to diver-based visual estimates or single camera systems [24][25][26]28]. This is advantageous for endangered species, such as sea turtles, which have until recently had limited behavioral studies in natural or artificial habitats (but see, e.g., [14,18,[29][30][31][32][33][34]). ...
Simple Summary
In-water behavioral observation studies of sea turtles are lacking primarily because they are highly migratory throughout their lifecycles. However, utilizing new technology, such as remote stereo-video cameras, allows researchers to collect size-specific data and view relatively natural sea turtle behavior. In this study, we completed stereo-video camera surveys at local artificial reefs, piers, and jetties within the northern Gulf of Mexico from May 2019 to August 2021. We remotely measured sea turtle straight carapace length, assessed wariness using minimum approach distance (MAD) between the turtle and camera operator, and documented behaviors during encounters. Our results indicated that green sea turtles had the smallest mean MAD before a startle response occurred, but the largest range compared to loggerhead and Kemp’s ridley sea turtles. Using a linear mixed model, we found that the size of the turtle was an important predictor of MAD. Overall, our results suggest that larger sea turtles became startled at longer distances compared to the smaller individuals surveyed across all the species studied. This study supports the use of stereo-video camera systems as a non-invasive tool to conduct surveys to assess sea turtle behavioral observations in relation to body size.
Abstract
Our understanding of size-specific sea turtle behavior has lagged due to methodological limitations. However, stereo-video cameras (SVC) are an in-water approach that can link body-size and allow for relatively undisturbed behavioral observations. In this study, we conducted SVC dive surveys at local artificial reefs, piers, and jetties in the northern Gulf of Mexico (nGOM) from May 2019 to August 2021. Using SVCs, we measured sea turtle straight carapace length, documented behaviors, and quantified wariness by assessing minimum approach distance (MAD). In green sea turtles (Chelonia mydas), the observed MAD ranged from 0.72 to 5.99 m (mean 2.10 m ± 1.10 standard deviation (SD), n = 73). For loggerhead sea turtles (Caretta caretta), the MAD ranged between 0.93 and 3.80 m (mean 2.12 m ± 0.99 SD, n = 16). Kemp’s ridley sea turtles (Lepidochelys kempii) were similar to loggerheads, and MAD ranged from 0.78 to 3.63 m (mean 2.35 m ± 0.99 SD, n = 8). We then evaluated what biological factors could impact the MAD observed by species, but we excluded Kemp’s ridleys as the sample size was small. Using a linear mixed model and model selection based on AICc, the top ranked model for both green and loggerhead sea turtles included SCL as the most important factor influencing MAD. MAD did not vary with habitat type for either species. Our results showed that larger individuals, regardless of species, have a greater wariness response, becoming startled at greater distances than smaller individuals. The findings of our study support the use of SVC as an accessible, non-invasive tool to conduct ecologically relevant in-water surveys of sea turtles to link behavioral observations to body size.
... As an alternative assessment method, stereo-camera surveys have been successfully applied to assess abundance and distribution and to conduct fish length measurements (Harvey et al., 2003;Watson et al., 2005;Shortis et al., 2009;Williams et al., 2016b;Boldt et al., 2018). Fish measurements obtained from stereo-camera imagery have been shown to be accurate (Harvey et al., 2003;Seiler et al., 2012). ...
... As an alternative assessment method, stereo-camera surveys have been successfully applied to assess abundance and distribution and to conduct fish length measurements (Harvey et al., 2003;Watson et al., 2005;Shortis et al., 2009;Williams et al., 2016b;Boldt et al., 2018). Fish measurements obtained from stereo-camera imagery have been shown to be accurate (Harvey et al., 2003;Seiler et al., 2012). In addition to developing estimates for fish abundance and morphological metrics, fish behavior can be observed (Somerton et al., 2017). ...
Forage fish and fish associated with particular benthic habitats (e.g., rockfishes, sand eels, sand lances) may be particularly difficult to assess through standard survey methodologies. Stereo-cameras, video, and automated visual data may serve as useful complementary tools to provide insight into the dynamics of these species. Visual methods may be used not only to estimate abundance and distribution, but also to inform important biological metrics and life history attributes. We explored the application of these methods to assess Pacific sand lance (Ammodytes personatus), a forage fish associated with benthic sediments, using a combination of directed observations from a manned submersible and quantitative analysis of fixed image footage obtained with a stereo-camera. This research provides a better understanding of how in situ observations and automated image analysis might complement other methods to estimate fish abundance, distribution, habitat, and behavior. Visual data were compared to data collected via directed sampling using physical extraction methods at the same site in the same year. Submersible observations provided new insights on the physical conditions and habitat. Visual observations confirmed wavefield morphologies previously identified through multibeam acoustic imagery and measured attributes relevant to the physical oceanography of the water column above this benthic habitat feature. Visual observations also informed understanding of light penetration, relevant to diurnal cues for seasonal progression and diel vertical migration and foraging. Submersible observations provided insights into abundance, schooling dynamics, and behavioral attributes, including avoidance in response to physical disturbance and aggregation in presence of artificial light. Quantitative analysis of stereo-camera data in center-edge and north-south transects determined that fish abundance and length distribution was relatively uniform throughout this particular benthic habitat. Estimates of measurement error associated with stereo-cameras were calculated and correction factors identified. Mean lengths estimated in visual data and in physical specimens were closely matched, though variance in visual data measurements was far greater. This error was reduced when filtering data on the basis of orthogonal position or incidence angle relative to the camera. Our research provides important insights to the presence, distribution, abundance, and movement of Pacific sand lance within benthic sand wavefield habitats. Our research also provides insight to the applications, opportunities, and constraints to observation-based sampling methods, including the use of manned submersibles and automated stereo-cameras.
... The pairs of cameras were mounted 0.7 m apart and angled in at 8 degrees to allow for stereo imaging. Filming in stereo adds the capability of making accurate measurements of individual fish, informing estimates of biomass (Harvey et al. 2003, Langlois et al. 2018. ...
This report communicates finding from a targeted monitoring program for the Discovery Bay Marine National Park in Victoria, Australia with the dual aims of understanding the condition of major conservations assets and identifying threats to those assets. Specific objectives of this project were to: • Compile seabed bathymetry data to create full-coverage maps for developing fish species distribution models. • Use baited remote underwater video station (BRUVS) surveys to assess fish populations, monitor demersal fish over time, evaluate no-take protection effectiveness, and analyse the impact of seafloor structure and oceanographic conditions on fish diversity using spatial distribution models. • Evaluate the abundance and size of Southern Rock Lobster inside and outside Discovery Bay MNP to determine protection effectiveness. • Utilize autonomous underwater vehicles (AUVs) to describe habitat types within and around the MPA.
... Early methods still relied on human labor. Harvey et al. (2003) developed a computer interface that utilizes cursor positioning and mouse clicks to locate the snout and caudal fin of fish, enabling the calculation of fish length based on the principle of stereoscopic intersection. The average error achieved was 1.72 millimeters. ...
The industrialization, high-density, and greener aquaculture requires a more precise and intelligent aquaculture management. Phenotypic and behavioral information of fish, which can reflect fish growth and welfare status, play a crucial role in aquaculture management. Stereo vision technology, which simulates parallax perception of the human eye, can obtain the three-dimensional phenotypic characteristics and movement trajectories of fish through different types of sensors. It can overcome the limitations in dealing with fish deformation, frequent occlusions and understanding three-dimension scenes compared to the traditional two-dimensional computer vision techniques. With the deep learning development and application in aquaculture, stereo vision has become a super computer vision technology that can provide more precise and interpretable information for intelligent aquaculture management, such as size estimation, counting and behavioral analysis of fish. Hence, it is very beneficial for researchers, managers, and entrepreneurs to possess a thorough comprehension about the fast-developing stereo vision technology for modern aquaculture. This study provides a critical review of relevant topics, including the four-layer application structure of stereo vision technology in aquaculture, various deep learning-based technologies used, and specific application scenarios. The review contributes to research development by identifying the current challenges and provide valuable suggestions for future research directions. This review can serve as a useful resource for developing future studies and applications of stereo vision technology in smart aquaculture, focusing on phenotype feature extraction and behavioral analysis of fish.
... and observing fish behaviour using stereo cameras 8,9 , and measuring the body length of tuna [10][11][12] . In addition, stereo cameras have been successfully used to measure fish body length in controlled aquaculture settings 13,14 and open water 15,16 . In general, stereo methods provide exact measurements compared to single-camera-based photogrammetric methods 17 . ...
Pacific saury is one of the most economically important species in the northwestern Pacific Ocean. The management of this resource relies on precise input of biological data such as body length and is often hindered by a lack of such data on captured fish. This study explores the potential of electronic monitoring (EM) using off-the-shelf stereo cameras to overcome the challenges of collecting and measuring saury body length from the Pacific saury fishery. Using a calibrated WEEVIEW SID WV3000 3D camera, a total of 252 paired images with different shooting angles and distances were obtained for further measurement using Sebastes Stereo Image Analysis Software (SSIAS). The treatments for the measurement distance (MD) were 30 cm, 60 cm, and 100 cm, for the depression angle (DA) were 30°, 60°, and 90°, and for the position angle (PA) were 0°, 45°, and 315° (− 45°). An assistant calibration (AC) in the form of a 16 cm black ruler was also added used. The SSIAS measurement results indicated that the best measurement was obtained with 0° position angle, 90° depression angle, and 30 cm distance from the target fish. The use of AC in the SSIAS + AC measurement was proven to reduce the measurement error from 2.45 to 8.64% to − 1.86 to 0.01%. This study set the baseline for the application of EM on collecting saury body length and the use of AC has been proven to increase the measurement accuracy.
... However, this approach imposes significant physical damage and physiological stress on the fish, necessitating capture and anesthesia for measurement. Consequently, recent research has focused on noncontact methods for measuring fish body size, such as acoustic measurement (Burwen et al. 2010;Cook et al. 2019;Folkedal et al. 2012;Knudsen et al. 2004;Puig-Pons et al. 2022) and optical stereo cameras (Harvey et al. 2003(Harvey et al. , 2002Komeyama et al. 2018;Muñoz-Benavent et al. 2018;Shafait et al. 2017;Tanaka et al. 2019aTanaka et al. , 2019bTorisawa et al. 2011Torisawa et al. , 2012. ...
To determine the optimal method for monitoring the size distribution of cultivated yellowtail growth, we employed three different approaches: capture measurement, manual measurement using stereo cameras, and automatic measurement through stereo camera-based image recognition technology. Conventional capture measurements showed inadequate prediction interval owing to limited sample size, preventing accurate assessment of growth. Both manual and automatic camera measurements successfully conformed to a growth model exhibiting periodicity. The expected values derived from each model closely matched with the mean of landings conducted at the end of the study. However, the 95% prediction interval for manual measurement with cameras was comparable to that for the landing measurement, whereas the prediction interval for the automatic measurement with cameras was overestimated. Additionally, the growth rate of farmed yellowtail demonstrated seasonal fluctuations. Notably, the mean obtained from a single automatic measurement with cameras, prior to landing, significantly deviated from the overall mean of all measurements. This suggests a potential risk associated with relying on accidental outliers in a single measurement. Therefore, it is crucial to employ a growth model unaffected by outliers in continuous measurements to ensure reliable predictions.
... Audio streams can be used to calculate temporal offset between two stereo cameras by cross correlating peaks in the waveforms of the two cameras' streams (Hasler et al., 2009). A flashing light, however, is preferably used to synchronise left and right camera images (Harvey et al., 2003) as audio can drift from video. We provide code for this (see Dunkley et al., 2023), which involves pinpointing at which frame (for each camera) the light starts flashing and calculating the difference between the two cameras. ...
Ecologists are now widely utilising video data to quantify the behaviours and interactions of animals in the wild. This process can be facilitated by collecting videos in stereo, which can provide information about animals' positions, movements and behaviours in three‐dimensions (3D). However, there are no published designs that can collect underwater 3D stereo data at high spatial and temporal resolutions for extended periods (days).
Here, we present complete hardware and software solutions for a long‐running, open‐source, underwater stereo camera rig, costing £1337. This stereo camera can continuously record aquatic species and their behaviours/interactions in high resolution (1080 p and 30 fps) and in 3D, over multiple days.
We provide full design guides for the cameras and a travel‐friendly rig, and include guidance and open‐source code for calibrating the cameras in space and time. We also show how these cameras could be used to track animals' body parts and positions, and how their size, posture and behaviour can be inferred.
This stereo camera will facilitate the collection of high‐resolution ecological and behavioural data, such as affiliative, agonistic or trophic interactions between species, which can inform us about the health and structure of ecosystems. These data will assist ecologists and conservationists in monitoring and understanding the impacts of current environmental pressures on ecosystem functioning.
... The pairs of cameras were mounted 0.7 m apart and angled in at 8 degrees to allow for stereo imaging. Filming in stereo adds the capability of making accurate measurements of individual fish, informing estimates of biomass (Harvey et al., 2003;Langlois et al., 2018). Each BRUVS frame was calibrated in a pool prior to fieldwork commencing. ...
Parks Victoria has established the Signs of Healthy Parks (SHP) program for its network of marine national parks and sanctuaries with the dual aims of improving baseline knowledge of Victoria’s marine protected areas (MPAs) and addressing applied management questions. The SHP program aims to monitor the health of protected areas using a range of environmental indicators that in turn provide information about the natural values and ecological processes within the parks, as well as potential threats and other drivers. This study implemented a suite of monitoring approaches to assess the health and condition of the no-take Wilsons Promontory Marine National Park (MNP) in eastern Victoria. The report details the finding of diver surveys, baited remote underwater video and habitat assessment to understand the long term trends, effectiveness of protection and habitat usage of key species of conservation interest inside the marine national park.
... For regulating the feeding process, stereo vision is significantly more advantageous in determining fish size and three-dimensional (3D) position. In recent years, stereo vision has been employed to count fish [31], predict fish biomass[32] conduct real-time classifications [33] measure fish size[34]. ...
The Power of AI in Fish Feeding and Behavior Management: A Comprehensive Review
Agniva Pradhan,
M.F.Sc (Dept. of Fish Nutrition, F.F.Sc, WBUAFS, KOLKATA)
Abstract: In this study, the use of computer vision and artificial intelligence technology is examined as a means of improving the efficacy and sustainability of fish feeding in aquaculture. The proposed approach employs machine learning techniques to estimate feeding requirements and optimise feeding schedules, while image analysis is employed to track fish behaviours and appetite. The method aims to increase the growth and health of fish while reducing waste and the environmental impact of aquaculture operations.
Keywords: Fish Feeding, Aquaculture, Sustainability, Optimization, Machine Learning, Artificial Intelligence, Computer Vision,
... RE is used to determine the error percentage of simulation in comparison to field measurement. RE is positive or negative according to whether the simulated values are an overestimate or an underestimate compared to observed values 73 . It is worth mentioning that the statistics of the model evaluation are affected by large errors and values which are extreme, especially in the case of small datasets. ...
The SIRMOD and WinSRFR models were used to model and assess the irrigation performance under continuous and surge irrigation strategies with two furrow lengths of 70 m and 90 m and stream sizes of 0.4 l/s and 0.6 l/s for each length. According to the normalized root mean squared error (NRMSE) and the relative error (RE), WinSRFR had, on average, excellent accuracy in the continuous and surge irrigation for simulating advance-recession times (NRMSE: 6.15 and 4.24% for advance time, and 2.20 and 5.20% for recession time), infiltrated water depth (NRMSE: 3.37 and 6.38%), and runoff volume (RE: 6.93 and 2.57%), respectively. SIRMOD had also, on average, excellent simulation in the continuous and surge irrigation for advance-recession times (NRMSE: 3.34 and 2.45% for advance time, and 2.28 and 6.41% for recession time), infiltrated water depth (NRMSE: 2.98 and 5.27%), and runoff volume (RE: 5.31 and 17.49%), respectively. The average of irrigation application efficiency (AE), distribution uniformity (DU), deep percolation (DP), and tail-water ratio (TWR) were 61.50, 90.25, 11.75, and 26.75% in continuous irrigation, and 72.03, 94.09, 8.39, and 19.57% in surge irrigation, respectively, which shows that surge irrigation increased AE (irrigation management performance) and DU (irrigation method performance) and reduced DP and TWR compared to continuous irrigation. Moreover, longer furrow lengths increased AE and DP under both irrigation methods, while it decreased TWR and DU. However, increasing the stream size decreased AE and DP and increased TWR under both continuous and surge irrigations. The higher stream size improved DU in continuous irrigation but reduced DU in surge irrigation. The results confirmed that both SIRMOD and WinSRFR are reliable analytical tools to evaluate furrow irrigation strategies for improving irrigation management. In conclusion, this study showed that surface irrigation models could be employed in practice by irrigation engineers and practitioners to design and define the optimized furrow length and stream size in arid and semi-arid areas where efficient and high performance irrigation strategies are required to save water and reduce water loss.
... Among other insights, they found that using a 3D calibration target, i.e., a wireframe cube, allowed for more precise measurements than calibration with a planar target, such as a checkerboard pattern. In Harvey et al. [39], stereo videos were utilized to measure free-swimming southern bluefin tuna. Specifically, the snout-to-fork length and the maximum body depth were measured with an average error of 1.72 and 1.37 mm, respectively. ...
The utilization of stationary underwater cameras is a modern and well-adapted approach to provide a continuous and cost-effective long-term solution to monitor underwater habitats of particular interest. A common goal of such monitoring systems is to gain better insight into the dynamics and condition of populations of various marine organisms, such as migratory or commercially relevant fish taxa. This paper describes a complete processing pipeline to automatically determine the abundance, type and estimate the size of biological taxa from stereoscopic video data captured by the stereo camera of a stationary Underwater Fish Observatory (UFO). A calibration of the recording system was carried out in situ and, afterward, validated using the synchronously recorded sonar data. The video data were recorded continuously for nearly one year in the Kiel Fjord, an inlet of the Baltic Sea in northern Germany. It shows underwater organisms in their natural behavior, as passive low-light cameras were used instead of active lighting to dampen attraction effects and allow for the least invasive recording possible. The recorded raw data are pre-filtered by an adaptive background estimation to extract sequences with activity, which are then processed by a deep detection network, i.e., Yolov5. This provides the location and type of organisms detected in each video frame of both cameras, which are used to calculate stereo correspondences following a basic matching scheme. In a subsequent step, the size and distance of the depicted organisms are approximated using the corner coordinates of the matched bounding boxes. The Yolov5 model employed in this study was trained on a novel dataset comprising 73,144 images and 92,899 bounding box annotations for 10 categories of marine animals. The model achieved a mean detection accuracy of 92.4%, a mean average precision (mAP) of 94.8% and an F1 score of 93%.
... The pairs of cameras were mounted 0.7 m apart and angled in at 8 degrees to allow for stereo imaging. Filming in stereo adds the capability of making accurate measurements of individual fish, informing estimates of biomass (Harvey et al. 2003, Langlois et al. 2018. Each BRUVS frame was calibrated in a pool prior to fieldwork commencing. ...
This study presents a suite of monitoring approaches implemented to assess the ecological health and condition of the no-take Cape Howe Marine National Park in the state of Victoria, Australia
... The other method is based on compensating for the refraction effects, in which the standard pinhole-camera model is used to describe the object-image relationship and the systematic refraction errors are additionally compensated in the calibration process [24]. Harvey and Shortis [25,26] performed calibration to compensate the refraction effect on the light path. However, there will always be some systematic errors that are not incorporated into the model. ...
Low-cost camera calibration is vital in air and underwater photogrammetric applications. However, various lens distortions and the underwater environment influence are difficult to be covered by a universal distortion compensation model, and the residual distortions may still remain after conventional calibration. In this paper, we propose a combined physical and mathematical camera calibration method for low-cost cameras, which can adapt to both in-air and underwater environments. The commonly used physical distortion models are integrated to describe the image distortions. The combination is a high-order polynomial, which can be considered as basis functions to successively approximate the image deformation from the point of view of mathematical approximation. The calibration process is repeated until certain criteria are met and the distortions are reduced to a minimum. At the end, several sets of distortion parameters and stable camera interior orientation (IO) parameters act as the final camera calibration results. The Canon and GoPro in-air calibration experiments show that GoPro owns distortions seven times larger than Canon. Most Canon distortions have been described with the Australis model, while most decentering distortions for GoPro still exist. Using the proposed method, all the Canon and GoPro distortions are decreased to close to 0 after four calibrations. Meanwhile, the stable camera IO parameters are obtained. The GoPro Hero 5 Black underwater calibration indicates that four sets of distortion parameters and stable camera IO parameters are obtained after four calibrations. The camera calibration results show a difference between the underwater environment and air owing to the refractive and asymmetric environment effects. In summary, the proposed method improves the accuracy compared with the conventional method, which could be a flexible way to calibrate low-cost cameras for high accurate in-air and underwater measurement and 3D modeling applications.
... Stereo video systems are a non-destructive method of counts and measurements that does not require the capture and handling of fish (Harvey et al., 2003) and an efficient tool for gathering information on the density, biomass and rate recovery of fish species (Langlois et al., 2006). Citizen science can increase the number of observations by voluntary divers in well-visited and accessible areas (Thiel et al., 2014) as well as on geographic scales difficult for researchers to access (Dickinson et al., 2010). ...
Here, the authors investigate the evolution of scientific literature on the Atlantic goliath grouper (Epinephelus itajara), the largest grouper species in the Atlantic. This species was considered threatened according to the IUCN for nearly three decades due to overfishing and has been protected from fishing in a large part of its range. For the purpose of evaluating the scientific production on an endangered species banned from fishing, the authors present a scientometry and synthesis review analysing the literature produced in the timeline and the content of their studies. To this end, they set up an almost 60‐year literary database through the Scopus, Web of Science and Google Scholar platforms. They analyse the publication and knowledge area patterns over the time in the world and in Brazil. They also feature the main areas, locations, ecosystems and types of those research studies and briefly describe the main records found in each decade. An increase in the number of publications was observed from the 1990s onwards worldwide, and particularly in Brazil from the 2000s, coinciding with the pioneering legislations for species protection. Most of the studies were classified within ecology and conservation and took place in countries that implemented moratoriums and had extensive areas of mangroves. Thus, fishing moratorium, classification as endangered, mangroves distribution and conservation all play a role in the studies distribution and contribute positively to the knowledge, as well as support conservation strategies for the species and its ecosystems of occurrence.
... To reduce the synchronization error, our video cameras recorded at 60 frames/sec. However, we recommend to record at higher (Neuswanger et al. 2016;Vivancos and Closs 2015) and SeaGIS (Bower et al. 2011;Harvey et al. 2003) showed that each software is able to estimate fish lengths with small variations. These variations may result from the straight-line distance between the fish's head and tail fork when swimming (Neuswanger et al. 2016). ...
In the last two decades underwater stereo-video analyses have received interest for use in diversity, abundance, and biomass estimations for marine species because of their low-cost, time-efficient and as a not invasive technique. Here, we assessed the accuracy and precision of measurements from two videogrammetric software (VidSync and EventMeasure by SeaGIS) from a Baited Remote Underwater Stereo-Video system (Stereo-BRUVs) for determining the size of objects of known length under controlled conditions and lengths of pelagic fish in open ocean conditions. One Stereo-BRUVs equipped with GoPro cameras was employed to take measurements of known-length rods in a pool to compare the influence of size, distance to rod and angle to the optical axis on measurement accuracy and precision. The ability to estimate fish length in open waters was tested by measuring 14 individuals of the common dolphinfish (Coryphaena hippurus) with the same Stereo-BRUVs equipment simultaneously calibrated for both software. The pool test showed that SeaGIS measurements (error = 0.22%) were more accurate than VidSync (error = 0.63%) only for the shortest rod. For both software, accuracy remained high (> 99%) and did not decrease regardless the size, distance, and angle of the measured rods. The open waters measurements showed high correlation (R² = 0.99) between VidSync and SeaGIS for the common dolphinfish, suggesting that both software take similar measurements of individual fish. We suggest both software make length estimations with great accuracy and precision.
... Research on semi-automatic methods mainly appeared in the early stages in the field of fish length measurement. Harvey et al. developed a procedure in the foundation of a stereo-vision system to measure the body length of tuna [15]. Hsieh et al. reported a semi-automated method using a calibration plate as a standard for dimensional correction by which a lack of scale in monocular images can be made up [16]. ...
Developing new methods to detect biomass information on freshwater fish in farm conditions enables the creation of decision bases for precision feeding. In this study, an approach based on Keypoints R-CNN is presented to identify species and measure length automatically using an underwater stereo vision system. To enhance the model’s robustness, stochastic enhancement is performed on image datasets. For further promotion of the features extraction capability of the backbone network, an attention module is integrated into the ResNeXt50 network. Concurrently, the feature pyramid network (FPN) is replaced by an improved path aggregation network (I-PANet) to achieve a greater fusion of effective feature maps. Compared to the original model, the mAP of the improved one in object and key point detection tasks increases by 4.55% and 2.38%, respectively, with a small increase in the number of model parameters. In addition, a new algorithm is introduced for matching the detection results of neural networks. On the foundation of the above contents, coordinates of head and tail points in stereo images as well as fish species can be obtained rapidly and accurately. A 3D reconstruction of the fish head and tail points is performed utilizing the calibration parameters and projection matrix of the stereo camera. The estimated length of the fish is acquired by calculating the Euclidean distance between two points. Finally, the precision of the proposed approach proved to be acceptable for five kinds of common freshwater fish. The accuracy of species identification exceeds 94%, and the relative errors of length measurement are less than 10%. In summary, this method can be utilized to help aquaculture farmers efficiently collect real-time information about fish length.
... Most of the fish size estimations systems (that are in use), employ multiple cameras. A high accuracy stereocamera was used for standard length estimation of southern bluefin tuna (Thunnus Maccoyii) [5] but the operator had to choose manually selected points from a video sequence. In another study [6], the accuracy assessment of a stereo camera was conducted by three experienced divers, estimating the length of reef fish on site. ...
... The two main kinds of fish size measuring methods based on stereo vision systems are size measurement based on manually selecting key points and size measurement based on automatically detecting key points of fish bodies. As the name implies, the method based on manual determination requires manually determining and situating the image's essential features, such as the positions of mouth or tail of fish, that are ideal for size measurement before using stereo vision to calculate the associated size parameter (Costa, et al., 2009;Harvey, et al., 2003;Rico-Díaz, et al., 2020;Williams, et al., 2016). These methods do not create pressure on the fish, but the process of determining key points still requires a lot of human participation. ...
Non-contact methods for estimating cultured fish weight are essential for aquaculture companies to develop aquaculture strategies and management plans. However, it is challenging for current technologies to precisely measure the size of fish in a densely breeding environment and estimate the weight of fish because of issues including occlusion, fish bending and not facing the camera. This study, which focuses on the aforementioned issues, suggests a technique for creating an instance segmentation dataset appropriate for measuring fish size as well as an attention-based fully convolutional instance segmentation network (CAM-Decoupled-SOLO) to extract fish contour features. This study uses a concatenation approach to fuse low-level features and high-level features in order to safeguard low-level features so as to address the issue of low-level feature information loss in the bottom-up propagation process of feature pyramid networks. Additionally, by combining pixel location information with channel attention mechanism, this work achieves the integration of target location information and channel dimension information. To validate the developed methodology, it has been applied to an aquaculture system. The results showed that the mAP of proposed method would be improved by 2.6% compared with Decoupled-SOLO. Both qualitative and quantitative assessments would be enhanced when compared to the existing well-liked one-stage network. Moreover, to save labor cost and avoid fish damage, an automatic fish perimeter measurement model and a weight prediction system were constructed by combining fish contours extracted by CAM-Decoupled-SOLO and binocular vision. There was a significant correlation between fish perimeter and weight, which can be used to estimate the weight of fish in complex aquaculture environments.
... To measure fish, several methods have been used over the past few years. These include machine learning approaches to improve identification [3], 3D measurement with a single fish-eye stereo camera [4] as well as double fish-eye [5], multiple fish-tracking algorithm [6], using paired images from 2 cameras [7] with a geometric algorithm to filter fish images [8], a classification algorithm with linear discriminant analysis tested by the leave-one-out cross-validation method [9], a practical system for measuring effective and simple fish size on very small fry accurately without injuring with contact less measurement consists of a USB camera mounted on a rod with an adjustable height and the camera will be connected to a computer for shooting [10]. This measurement method is divided into 2 parts, namely: contact methods and non-contact methods [11]. ...
... Industrial facilities, such as offshore wind turbine foundations, oil and gas pipeline systems, underwater structures, and other objects, including anchor chains, are regularly required to be measured underwater [1,2]. Other objects of interest for underwater 3D measurements include archaeological sites on the seabed [3][4][5][6][7], sunken shipwrecks [8], plant and coral growth [9], and the biomass or size of fish populations [10][11][12]. Several contactless methods have been applied to perform 3D reconstruction underwater. ...
Featured Application
A potential application of the work is the underwater 3D inspection of industrial structures, such as oil and gas pipelines, offshore wind turbine foundations, or anchor chains.
Abstract
A new underwater 3D scanning device based on structured illumination and designed for continuous capture of object data in motion for deep sea inspection applications is introduced. The sensor permanently captures 3D data of the inspected surface and generates a 3D surface model in real time. Sensor velocities up to 0.7 m/s are directly compensated while capturing camera images for the 3D reconstruction pipeline. The accuracy results of static measurements of special specimens in a water basin with clear water show the high accuracy potential of the scanner in the sub-millimeter range. Measurement examples with a moving sensor show the significance of the proposed motion compensation and the ability to generate a 3D model by merging individual scans. Future application tests in offshore environments will show the practical potential of the sensor for the desired inspection tasks.
... There are many fields of application in archaeology, marine industry, offshore energy production, fishing industry, and environmental tasks. Examples are documentation of underwater historical archaeological sites (Roman et al. 2010;Menna et al. 2018), sunken shipwreck exploration (Korduan et al. 2003), size and volume estimation of fish (Harvey et al. 2003;Dunbrack 2006;Costa et al. 2006), time dependent coral reef growth (Bythell et al. 2001;Guo et al. 2016), and inspection (Tetlow and Spours 1999;McLeod et al. 2014) of oil or gas pipelines, industrial structures, or offshore wind farm foundations. ...
A novel concept of camera modelling for underwater 3D measurements based on stereo camera utilisation is introduced. The geometrical description of the ray course subject to refraction in underwater cameras is presented under assumption of conditions, which are typically satisfied or can be achieved approximately. Possibilities of simplification are shown, which allow an approximation of the ray course by classical pinhole modelling. It is shown how the expected measurement errors can be estimated, as well as its influence on the expected 3D measurement result. Final processing of the 3D measurement data according to the requirements regarding accuracy is performed using several kinds of refinement. For example, calibration parameters can be refined, or systematic errors can be decreased by subsequent compensation by suitable error correction functions. Experimental data of simulations and real measurements obtained by two different underwater 3D scanners are presented and discussed. If inverse image magnification is larger than about one hundred, remaining errors caused by refraction effects can be usually neglected and the classical pinhole model can be used for stereo camera-based underwater 3D measurement systems.
... CV = 0.30%). This accuracy is comparable to that of many photogrammetric systems used to measure marine wildlife (Harvey et al. 2003;Growcott et al. 2012;Letessier et al. 2013;Boutros et al. 2015;Durban et al. 2015;Dawson et al. 2017;May et al. 2019) indicating that the approaches used in calibration and distortion correction were successful. The greater variation seen in the five replicate measurements of each shark (mean CV = 2.1%) was likely the result of variation in shark body curvature during measurement, and human error in precisely locating the measurement points. ...
The broadnose sevengill shark (Notorynchus cepedianus) is a large marine predator found in temperate coastal marine habitats. Despite being commonly encountered, details of its ecology are limited. To partially address this, we quantified the size and sex structure of N. cepedianus in Paterson Inlet, Stewart Island, New Zealand. A stereo-camera baited underwater video system (BRUV) was deployed in a shallow coastal embayment to capture underwater video of N. cepedianus for stereo photogrammetric measurement between June 2016 and November 2017. Individual sharks were sexed via observation of claspers. Three-dimensional underwater calibration of the photogrammetric system, using a precisely measured cube, indicated a mean error of <1%. Estimated total lengths ranged from 149.0 to 276.5 cm (n = 78 individuals, mean = 211.4, S.E. = 2.97). The observed sex ratio was heavily biased, with females making up 95% of the population in winter 2016 and 73% in summer 2017. This study provides new data on N. cepedianus in southern New Zealand as well a non-invasive, low-cost methodology that can be applied to other study areas.
... An underwater stereo-video system mounted on the ROV was chosen to record fish, as this system allowed an accurate measurement of fish length (Harvey and Shortis, 1995;Harvey et al., 2002bHarvey et al., , 2010 to within 1% of the true length (Harvey et al., 2002a(Harvey et al., , 2003. The high level of accuracy and precision of the measurements facilitated the conversion of length estimates to biomass using available length/weight regressions (Harvey et al., 2002a). ...
Decommissioning of offshore oil and gas structures is either occurring, or imminent in most regions of the world. Most jurisdictions require that offshore structures be removed for onshore disposal. However, there is growing interest in understanding the ecological and socio-economic benefits of leaving structures in the water. Descriptions of how fish utilize the vertical structure created by wellhead platform jackets (platforms) will provide insights into possible outcomes of decommissioning alternatives, such as full removal, leave in situ, or translocation to a designated reefing site. We surveyed fish assemblages associated with seven platforms and five reference sites located ∼150 km offshore in the central Gulf of Thailand. The platforms spanned the entire water column (∼75 m) and were a mix of three and four legged structures. We used a Remotely Operated Vehicle (ROV) fitted with an underwater stereo video system to quantify the abundance, size, biomass, and economic value of fish associated with the platforms. We recorded 43 species of fish on the platforms and five reference sites with most fishes on platforms categorized as coral-reef or coral-reef-associated species. We observed a strong vertical zonation in the fish assemblage on the platforms. The Regal demoiselle (Neopomacentrus cyanomos) was numerically dominant (75% of all fish observed). We measured 3,933 kg of fish on the platforms with Caranx sexfasciatus accounting for 76.12% of that. We conservatively estimate each platform had a scaled mean biomass of ∼2,927 kg and the fished species had scaled mean economic value of 175,500 Thai Baht per platform. We estimated that the biomass of fish associated with the seven platforms was at least four times higher per unit area than some of the world’s most productive coral reefs.
Real-time estimation of fish growth offers multiple benefits in indoor aquaculture, including reduced labor, lower operational costs, improved feeding efficiency, and optimized harvesting schedules. This study presents a low-cost, vision-based method for estimating the body length and weight of olive flounder (Paralichthys olivaceus) in tank environments. A 5 × 5 cm reference grid is placed on the tank bottom, and images are captured using two fixed-position RGB smartphone cameras. Pixel measurements from the images are converted into millimeters using a calibrated pixel-to-length relationship. The system calculates fish length by detecting contour extremities and applying Lagrange interpolation. Based on the estimated length, body weight is derived using a power regression model. Accuracy was validated using both manual length measurements and Bland–Altman analysis, which indicated a mean bias of −0.007 cm and 95% limits of agreement from −0.475 to +0.462 cm, confirming consistent agreement between methods. The mean absolute error (MAE) and mean squared error (MSE) were 0.11 cm and 0.025 cm2, respectively. While optimized for benthic species such as olive flounder, this system is not suitable for free-swimming species. Overall, it provides a practical and scalable approach for non-invasive monitoring of fish growth in commercial indoor aquaculture.
There is global awareness that many species of elasmobranchs (sharks and rays) have life history characteristics that make them susceptible to overexploitation. The study of these animals is critical, as it contributes to increasing knowledge of these specimens and aids in their conservation. In particular, growth rate, age, fecundity, and size at maturity are key parameters for defining management and conservation strategies in elasmobranchs. Biometric data collection allows these parameters to be determined and considered in the evaluation of population demography. Over the last decades, several methodologies for measuring elasmobranch size have evolved, progressing from traditional capture-based methods to sophisticated, non-intrusive photographic techniques. The present review aims to understand and analyse all the existing non-invasive techniques that currently allow the collection of zoometric data in elasmobranchs and, later, to highlight the advantages and limitations of each technique, with comments on their application to fieldwork. To this end, 49 articles were selected, encompassing seven measurement techniques: photogrammetry using distance to the individual, bar photogrammetry, laser photogrammetry, stereo-DOV, stereo-BRUV, stereo-ROV, and aerial photogrammetry. Globally, the last four techniques are excellent alternatives to methods that involve animal capture or death, as they are practical, simple to use, minimally invasive, and potentially highly accurate. Each technique's requirements related to equipment and cost, limitations, and distinctive features are presented here and summarized to guide researchers on what's available and how to select the most appropriate for their studies.
This study exploits a convenient, stable, accurate estimation of the spatio-temporal distribution of ocean wave heights by stereo observation using a single unmanned aerial vehicle (UAV) with two independent optical cameras fixed to it. To accurately estimate the spatio-temporal distribution of water levels, the photographing rate of each camera is set to 240 frames per second (fps), and luminescence captured during flight is used to realize the time synchronization within ~ 4 ms between the cameras. Based on the setting of the UAV and the cameras, we carry out aerial stereo observation of the spatio-temporal distribution of nearshore water levels. The estimated water levels showed good agreement with ground truth observation. We examine the dependence of the estimation accuracy on the photographing time difference between the cameras. Simulations were conducted to increase the time difference by decreasing the photographing rate. The estimation based on low photographing rates (e.g., 60 and 30 fps) likely failed when breaking wave crests prevailed in the optical images. This indicates that strict time synchronization (e.g., ~ 4 ms) between the cameras is an important, necessary condition to accomplish accurate stereo observation of ocean waves including wave breaking.
The effective implementation of machine vision has played a crucial role in advancing intelligent aquaculture across various domains. Stereo vision, as a branch of machine vision, has become a mainstream technology in aquaculture. Its distinctive capability to conduct comprehensive underwater monitoring from multiple angles, unaffected by object occlusion has propelled it to the forefront of aquaculture applications. This article offers a comprehensive review of the diverse applications of stereo vision in aquaculture spanning from its inception to present. The exploration encompasses its role in crucial areas such as biomass estimation and behavioural analysis, which include fish counting, weight estimation, swimming behaviour, feeding behaviour and abnormal behaviour. Furthermore, the paper delves into the advantages of stereo vision over traditional 2D machine vision approaches, while also acknowledging limitations, and identifying future challenges that must be addressed to fully leverage its potential in aquaculture. The review emphasizes the prospect of advancement in deep learning stereo‐matching algorithms specifically designed for underwater environments to catalyse a breakthrough in stereo vision technology. In summary, this review aims to provide researchers and practitioners with a better understanding of the current development of stereo vision in aquaculture, optimizing stereo vision technology and better serving the aquaculture field.
Reef fish assemblages across tropical and temperate latitudes are increasingly threatened by human impacts and climate change. Accurate and efficient survey methods are essential for quantifying these communities to inform management strategies. Imaging sonars (ISs) are high-frequency acoustic devices that produce camera-like images of objects. Unlike optical instruments, IS functions effectively in turbid and dark water and has proven valuable in detecting fishes in poor visibility and at night. Abundance, species richness, and fish size are desirable metrics in most reef fish surveys. This review investigates previous attempts to quantify these metrics using IS across different habitats. These metrics are often quantified in comparison with alternative methods (e.g. cameras, extractive techniques). This review examines the causes of agreement or incongruence between estimates from IS and estimates from these alternatives. Any instrument employed to quantify reef fishes should be able to operate in structurally complex habitats, and thus the ability of IS to function in these circumstances is also reviewed. Finally, 5 notable limitations of IS are described and solutions discussed. Overall, this review underlines the net value of IS for surveying reef fishes but advises using alternative methods to complement IS estimates of abundance, species richness, and fish size.
Assessing the health of fish populations relies on determining the length of fish in sample species subsets, in conjunction with other key ecosystem markers; thereby, inferring overall health of communities. Despite attempts to use artificial intelligence (AI) to measure fish, most measurement remains a manual process, often necessitating fish being removed from the water. Overcoming this limitation and potentially harmful intervention by measuring fish without disturbance in their natural habitat would greatly enhance and expedite the process. Stereo baited remote underwater video systems (stereo-BRUVS) are widely used as a non-invasive, stressless method for manually counting and measuring fish in aquaculture, fisheries and conservation management. However, the application of deep learning (DL) to stereo-BRUVS image processing is showing encouraging progress towards replacing the manual and labour-intensive task of precisely locating the heads and tails of fish with computer-vision-based algorithms. Here, we present a generalised, semi-automated method for measuring the length of fish using DL with near-human accuracy for numerous species of fish. Additionally, we combine the DL method with a highly precise stereo-BRUVS calibration method, which uses calibration cubes to ensure precision within a few millimetres in calculated lengths. In a human versus DL comparison of accuracy, we show that, although DL commonly slightly over-estimates or under-estimates length, with enough repeated measurements, the two values average and converge to the same length, demonstrated by a Pearson correlation coefficient (r) of 0.99 for n=3954 measurement in ‘out-of-sample’ test data. We demonstrate, through the inclusion of visual examples of stereo-BRUVS scenes, the accuracy of this approach. The head-to-tail measurement method presented here builds on, and advances, previously published object detection for stereo-BRUVS. Furthermore, by replacing the manual process of four careful mouse clicks on the screen to precisely locate the head and tail of a fish in two images, with two fast clicks anywhere on that fish in those two images, a significant reduction in image processing and analysis time is expected. By reducing analysis times, more images can be processed; thereby, increasing the amount of data available for environmental reporting and decision making.
The present study assessment of energy use, economic productivity and environmental life cycle assessment (LCA) for different production of fish systems in Alborz province. Productivity indicators and the provision of necessary solutions in order to improve the values of energy and economic indices in fish breeding ponds have been implemented. For this purpose, information related to the two fish systems sites was collected by a questionnaire and direct reference to the site owners and administrators and reference to the records. Then, the amount of energy input and output of cold-water and hot-water fish was calculated. The results showed energy ratio (ER) (2.24), energy productivity (EP) (0.04 kg MJ−1), energy intensity (EI) (26.83 MJ kg−1) and net energy gain (NEG) (33222.16 MJ kg−1) of cold-water fish were reported to be more favorable. According to these results, among energy use inputs, fish feed and electricity are two factors that proper management of their consumption increases energy efficiency. The results of economic analysis showed the benefit-to-cost ratio was positive for cold-water fish (1.54) and hot-water fish (2.45). Productivity for cold-water and hot-water fish was 0.58 Kg −1, respectively. LCA results showed that damage assessment of human health, ecosystem quality and resources for cold-Water and hot-water fish production are (0.00041 and 0.00045) DALY, (4.04E-08 and 4.61E-08) species.yr and (0.076 and 0.087) USD2013, respectively. The results of cumulative exergy demand (CExD) also indicated that the energy form of non-renewable, fossil is significant for both types of fish.
Information on the length of cultured fish is crucial to evaluate their growth status and biomass. However, existing methods to estimate fish body length are based primarily on manual sampling, which is invasive, time-consuming, and labor-intensive. In contrast, such information can be acquired via noncontact methods relying on images of the animals’ centerlines in an image. The centerline extraction method has been widely applied as an efficient approach. However, the greatest challenge in measuring the length of cultured fish is that of accurately extracting the animals’ centerlines when fish overlap in an image. To address this problem, we propose a method to obtain the centerlines of overlapping fish by processing images captured from a top-view perspective. First, fish contours are segmented using convex-concave points. Second, the contour segmentations belonging to each fish are identified by an algorithm that matches heads to tails; their centerlines are thus extracted and optimized. We conducted both laboratory and on-site experiments, and results show that the proposed method exhibited excellent performance in extracting fish centerline information, even for fish that overlapped in images captured from a top view. Moreover, it also improved on the performance of existing methods in measuring the length of cultured fish from such images.
The recent commercialization of unoccupied aerial vehicles (UAVs) has facilitated their incorporation into a variety of ecological studies. While UAVs are able to provide accurate visual data of marine species from an aerial perspective, these devices have some limitations that make measuring marine animals below the surface challenging. Many marine organisms are often visible from the air, but are deeper in the water column, and current methods cannot measure animals below the surface. Here, we developed and tested a stereo-video camera (SVC) system that was mounted onto a commercially-available UAV. We used the SVC-UAV to conduct remote body-size measurements for two marine species: the green sea turtle (Chelonia mydas) and the nurse shark (Ginglymostoma cirratum). When comparing SVC measurements to those taken by hand, the SVC-UAV had a mean absolute error (MAE) of 4.44 cm (n = 6; mean percent error (MPE) = 10.6%) for green sea turtles and 7.16 cm absolute error (n = 1; PE = 3.6%) for the nurse shark. Using a linear model, we estimated the slope of the SVC versus hand measurements for green sea turtles to be 1.085 (±0.099 SE), and accounting for the standard error, a measurement bias was not apparent. Using model selection, based on a global model predicting MAE from animal distance to the SVC and body size, the top ranked model was the intercept-only model. This indicates that neither animal distance nor body size strongly influenced measurement error. Incorporating SVC systems into UAVs can allow for relatively accurate measurements of near surface-dwelling marine species. To our knowledge, there is no other stand-alone SVC for UAVs available that offers similar accuracy and utility.
Since field surveys of fish school behavior around fishing gear are difficult to conduct, simulation of such behavior is of great potential significance with regard to the capture process of the fishing gear, especially in case of rare and endangered species of fish. To obtain model parameters for the school behavior of juvenile Pacific bluefin tuna (jPBT) in seawater, a portable stereo camera was used to measure in three dimensions the nearest neighbor distances of five jPBTs and distances between the net and the fish school in a fish net cage at sea. The results of the performance test of the stereo camera showed that the measured distances were 100.6% and 101.9% of the actual measured lengths at distances of 36 cm and 100 cm from the camera, respectively. We obtained five continuous video frames in which all jPBT were pictured within the angle of view. The nearest neighbor distance and the distance between the centroid of the fish school and the side net were 17.6±5.36 cm and 28.0±0.26 cm, respectively. Although a measurement error of less than 2% was observed, this is not significant as compared with the errors in previous studies; therefore, the portable stereo camera is effective for monitoring fish schools.
Fish biomass is one of the reliable parameters that can provide insight into fish and environmental health. Estimation of biomass in a dense and occulted environment is an inevitable and challenging task in modern aquaculture industries, which has been addressed in this work. The proposed work aims to determine the length features of fish using a deep learning-based segmental analysis technique. It tends to analyze the visibility of fish segments like head, body, and tail to define a completely visible fish (CVF). YOLOv4 (You look only once – Version-4) deep learning model is trained and used to detect the fish head, body, and tail segments. The detected segments are associated using sequence constrained nearest neighborhood (NN) association technique guided with fish head orientation. Fish length is estimated using the measurement points identified in the CVF. The measurement point includes head-start, body-center, and tail-end points, which are identified using a convex hull and oriented bounding box (BB). A calibration curve expressing the length-mass relation is used to determine the fish biomass from the estimated length. The proposed methodology is applied to determine the biomass of the genetically improved farmed tilapia (GIFT) fishes in an occulted environment. Experimental results illustrate a 0.9451 mAP of the trained YOLOv4 model and about 95.4% CVFs are detected accurately. A reliable accuracy of 94.15% and 91.52% is observed with testing and validation image sets respectively for biomass estimation.
Visual census conducted by divers has remained the most common method for obtaining quantitative data on reef fish assemblages since it was first applied in the 1950s. More recently, stereo-video systems are increasingly becoming popular as a sampling method to complement underwater visual counts. Understanding which method is more suitable to estimate fish assemblage metrics for particular objectives and hypotheses is important to design effective studies. Here, we compared rocky reef fish assemblage characteristics sampled by diver-based stationary point count and transect visual census, as well as back and forth diver-operated stereo-video systems (stereo-DOVs) and baited remote underwater stereo-video systems (stereo-BRUVs) in a coastal island in Southwestern Atlantic. Diver-based methods sampled a higher richness and abundance of fishery non-target species closely associated with the reef, as well as cryptobenthic small-bodied species. Conversely, stereo-BRUVs’ samples focused more on larger and mobile carnivorous fishery target species of the families Carangidae and Lutjanidae. The stereo-DOV recorded higher species richness compared to other methods and required less sample effort in the field. Limitations faced by the four methods were discussed; however, the higher sampling efficiency of the adapted back and forth stereo-DOVs, combined with the diverse advantages of video techniques, such as permanent record of data and decreased interobserver variability for species identification and estimates of body size, indicates stereo-DOVs as a highly efficient method to assess the entire fish assemblage, across feeding guilds and commercially important species. In addition, when combined with a remote technique, such as stereo-BRUVs, both of these methods efficiently capture estimates of diversity metrics as well as body size information especially for fishery target species in rocky reefs.
Age and growth of southern bluefin tuna (SBT, Thunnus maccoyii ) were estimated based on otoliths collected by Korean tuna longline vessels in the Atlantic and Indian Oceans from 2015 to 2019. A total of 739 specimens were used to estimate age and growth of SBT, with sizes ranging from 66 to 181 cm in fork length. It was confirmed that the otolith is a suitable aging characteristic for determining SBT age, and otolith annuli can be used as an annual ring. The relationship between fork length (FL) and total weight ( W ) was W = 7.7e − 05FL 2.722 ( R ² = 0.874). The von Bertalanffy’s growth parameters estimated from the non-linear method using length-at-age data were L ∞ = 170.0 cm, K = 0.200/year, t 0 = − 1.615 years, with 95% confidence intervals between 166 and 175 cm for L ∞ , 0.17–0.23/year for K , and − 2.27 to − 1.09 years for t 0 from bootstrapping. The Kimura’s likelihood ratio test results of the models under assumptions of common and different parameters between female and male concluded that the growth of SBT has no difference between genders.
The sedimentological processes taking place in a river massively affect the design, construction and operation of structures situated beside or in the river. Existing sedimentological models, which emerged from idealized physical model experiments, often fail to adequately describe the behaviour of the riverbed in the vicinity of structures. However, the improvement of existing sedimentological models as well as the design of new approaches require more accurate measurement data. The measurement data should resolve the transport process at grain scale with a high temporal resolution. Existing measurement methods do not deliver sufficient spatially and temporally resolved sedimentological data, as they specialize either in a high spatial resolution or in a high temporal resolution. The measurement system developed in this thesis provides both a high spatial and temporal resolution and thus provides a deep insight into riverbed morphological processes. In this thesis, I develop an underwater photogrammetric system from which a far more accurate digital model of the riverbed can be generated, providing deeper insight into the complex morphological processes taking place in a movable riverbed. A multi-camera system equipped with hemispherical lenses took the underwater images. The image acquisition took place underwater while the experiment was running, which prevents unwanted interruptions. The system recorded the dynamic processes of the riverbed in high spatial and temporal resolution. Together with the developed processing technique, this allows high resolute point clouds to be generated on a single grain scale level. The Introduction gives an overview of various methods for measuring morphological changes and briefly discusses their suitability for accurate modelling. With the aim of recording the riverbed topography with high spatial and temporal resolution on single scale level, photogrammetry has emerged in the field as the most suitable means of obtaining refined measurements. In the further course of the thesis, I discuss the photogrammetric and optical peculiarities resulting from the application in a multi-media environment. In the Photogrammetry and Optics sections, which deal with the theoretical background, the focus is on the distance dependence of the underwater distortions, the distortion characteristics of the hemispherical lenses, and their optical imaging behavior in multi-media applications. The practical tests were carried out at the Obernach Hydraulics Laboratory of the Technical University of Munich, where the multi-camera system was installed in a channel with a movable bed and tested extensively. Using a special procedure for center alignment, which establishes the coincidence of the entrance pupil of the lenses with center to hemispherical lens within a few tenths of a millimeter, it was possible to limit the distortion caused by the hemispherical lenses to a few pixels. The remaining residual distortions of the hemispherical lens were quantified by means of an underwater calibration, and removed from the underwater images for a highly accurate reconstruction of the riverbed topography. By conducting image correction to remove refractions from the image as outlined in this thesis, the systematic errors normally produced by conventional photogrammetric modelling can be expunged. Further processing steps create spatially triangulated meshes, digital terrain models and, together with the images, orthophotos of the bed. With these images and survey data, the riverbed can be described digitally. The extended outlook presents the potential of the developed measuring system for obtaining information relevant to hydraulic engineers, including roughness, mean riverbed elevations and single grain movements. Future research can make use of such data to improve existing sedimentological models and to develop novel approaches, such as data-driven models. In addition, it would be useful to combine the developed measuring system with other measuring methods, for example to measure flow velocities simultaneously.
Visual estimates of reef fish length are a non destructive and useful way of determining the biomass,mean length, or length frequency of reef fish. Consequently, visual estimates of
reef-fish length are often an important component of reef-fish monitoring programs,
many of which increasingly use volunteers. We compared estimates of the length of plastic fish silhouettes determined visually by experienced scientific and novice SCUBA divers. Novice divers showed a similar level of accuracy (mean error: 2.3 cm) to that of experienced
scientific divers (mean error: 2.1 cm). Significant improvements in accuracy and precision were provided by a stereo-video system (mean error: 0.6 cm). After minimal training in the
use of hardware and software, volunteers can obtain a high degree of measurement accuracy and precision with a stereo-video system, allowing them to assist with monitoring reef-fish lengths.
Underwater tests using plastic silhouettes of fish were used to compare the accuracy and precision of measurements made with a single video camera system to those made from two stereo-video systems (one using digital camcorders, the other using Hi8 camcorders). Test measurements made across a variety of ranges and angles of silhouette orientation in the fields of view showed the length estimates from both the digital and Hi8 stereo-video systems were substantially more accurate and precise than those obtained by the single video camera system, and had the great advantage that the position (range and bearing) and orientation of a fish target could be measured directly. Measurements made with stereo-video were much less restricted by range and subject orientation than those made with single video. The data resulting from these trials are used to propose a set of guidelines to optimize the accuracy and precision of underwater measurements of fish length using single and stereo-video systems.
Assessment of population structure is often used to detect and analyse the impact on marine fauna of environmental factors or commercial fishing. A primary tool in the characterisation of population structure is the distribution of the lengths of a large sample of individual specimens of a particular species. Rather than use relatively unreliable visual estimates by SCUBA divers, an under-water stereo video system has been developed to improve the accuracy of the measurement of lengths of highly indicative species such as reef fish. In common with any system used for accurate measurements, calibration of the camera system is of paramount importance to realise the maximum possible accuracy from the system. Further, the determination of the relative orientation of the two cameras is vital to the correct estimation of fish lengths. Also at issue is the stability of the calibrations and relative orientation of the cameras during deployments to capture video sequences of marine life, as any variations will inevitably lead to systematic errors and therefore inaccuracies in the measured lengths. This paper describes a series of experiments concerning the determination and testing of camera calibration, relative orientation and stability of the underwater stereo video system. The strategy for the integrated determination of calibration and orientation is described. Variations in calibration and orientation parameters are quantified in terms of magnitude and significance. Finally, the detected variations are analysed for the propagated effects on object space accuracy.
Assessment of age and size structure of marine populations is often used to detect and determine the effect of natural and anthropogenic factors, such as commercial fishing, upon marine communities. A primary tool in the characterisation of population structure is the distribution of the lengths or biomass of a large sample of individual specimens of a particular species. Rather than use relatively unreliable visual estimates by divers, an underwater stereo-video system has been developed to improve the accuracy of the measurement of lengths of highly indicative species such as reef fish. In common with any system used for accurate measurements, the design and calibration of the underwater stereo-video system are of paramount importance to realise the maximum possible accuracy from the system. Aspects of the design of the system, the calibration procedure and algorithm, the determination of the relative orientation of the two cameras, stereo-measurement and stereo-matching, and the tracking of individual specimens are discussed. Also addressed is the stability of the calibrations and relative orientation of the cameras during dives to capture video sequences of marine life.
The in situ estimation of the length of marine biological specimens or features by Scuba divers is complicated by the magnifying properties of water and the underwater environment. In this paper we describe the development and preliminary testing of an underwater stereo-video system which is designed to make accurate and precise measurements of the lengths of marine subjects. The results of 640 measurements of 16 plastic silhouettes of fish are presented. The initial results demonstrate the ability of the system to make accurate and repeatable measurements of length when the orientation of the subject to the stereo-video cameras is less than 75 degrees.
We calculated the power of visual length estimates by novice and experienced scientific SCUBA divers and estimates generated by a stereo-video system to detect changes in the mean length of three common species of reef fish from New Zealand. Length estimates from a stereo-video system had much greater power for blue cod (mean length=33.1 cm., range 19.5-50.1 cm.) and snapper (mean length=31.7 cm., range 23-71 cm.). For a third species, red cod (mean length=42.5 cm., range 13-74 cm.), the statistical power of diver and stereo-video estimates was much less for an equivalent number of samples owing to the greater variation in the true mean length of red cod recorded at different sites. At 90% power, a stereo-video system detected a 15% (similar to5-cm) change in the mean length of blue cod with 63% less samples (10) than those required by the experienced scientific divers (27). Novice scientific divers required 28 samples.
A video-based system which can be used to measure the average fish size in a nonintrusive fashion was developed. It performed well under a wide range of fish behaviours, types, stocking densities, ages and water quality conditions in tanks and sea cages. The design was based on principles of simple stereo geometry, incorporated fish dimensions/weight relationships and took into consideration fish movement to lower system costs. The overall system is durable, cost effective and easy to use. No artificial lighting is required. Approximately 20 min of continuous video footage are needed, and depending on the fish-image density, 30–60 min of labour are required to select suitable fish images and determine accurately the average fish size of a population. On still fish, average fish mass was measured to within 0·5%. On moving fish in sea cages and tanks with known individual salmonid masses, no significant difference was evident between measured and actual average mass. Testing the video system against the crowding/dip netting method in trout hatchery tanks containing from 7000 to 20 000 fry showed the video system to be as accurate as the crowding/dip netting method for determining fish mass provided the cameras were positioned in apparently preferred fish swimming locations. The new video method can be used to economically record and study behaviour, swimming speed and size distributions under a wide range of culture conditions.
Using simultaneous measurements from an underwater stereo-video system the accuracy and precision of length estimates of reef fish made by three experienced diver scientists under field conditions is determined. The trial showed that under optimal conditions the divers’ estimates were very accurate but lacked precision . The effects of the low precision are then demonstrated by using these field estimates to model the theoretical statistical power of the scientific divers to detect changes in the mean length of three species of fish from New Zealand coastal waters. The results suggest that the experienced diver scientists would have a much lower statistical power than stereo-video measurements to detect changes in the mean length, especially where low numbers of fish are recorded or where a research program aims to detect a change in the mean length of a fish population of 30% or less.
A photogrammetric technique is described that can be applied to
the problem of measuring the linear dimensions of free-swimming salmon
contained in aquaculture pens. These dimensions are then used to
statistically estimate the animal's weight. The system is non-invasive
and is small and logistically easy to deploy from small boats. The
system employs a very simple stereoscopic calculation that is
ratio-metric. This means that image measurements are made in arbitrary
units and converted to physical units by a multiplicative constant. This
constant is simply the inter-pupil distance between the lenses of the
stereoscopic pair of video cameras. Additionally, by choosing images in
which the target is swimming perpendicular to the optical axis of the
apparatus, the calculation is independent of the distance from the
cameras to the fish, although full-frame images produce the greatest
accuracy. The video cameras are frame synchronized for greater noise
immunity and to ensure image-pair simultaneity. That is, the signals
representing the 30 frames per second produced by each video camera are
coincident in time. When combined they are perfectly matched. The usual
practice would be to record each video signal using separate VCRs with
synchronized time-coding for image identification and matching. This
would be cumbersome and subject to operator error. The system under
discussion avoids these difficulties by using an electronic device to
“split” each of the two signals and to record the useful
parts of each as a combined image using a single VCR
A comparison of the accuracy and precision of digital and anal-ogue stereo–video systems
- E S Harvey
- M R Shortis
- M Stadler
- M Cappo
Harvey, E.S., Shortis, M.R., Stadler, M., Cappo, M., 2002b. A comparison of the accuracy and precision of digital and anal-ogue stereo–video systems. Mar. Tech. Soc. J. 36 (2), 38–49.
VMS Reference and User Guide
- M R Shortis
- S Robson
Shortis, M.R., Robson, S., 2001. VMS Reference and User Guide. http://www.geomsoft.com.au.
Southern bluefin tuna fishery
- C Robins
- T Polacheck
- A Preece
- A Caton
Robins, C., Polacheck, T., Preece, A., Caton, A., 2000. Southern bluefin tuna fishery. In: Caton, A., McLoughlin, K. (Eds.), Fishery Status Reports 1999: Resource Assessments of Australian Commonwealth Fisheries. Bureau of Rural Sciences, Canberra, Australia, pp. 159–168.
An underwater digital photogrammetric system for fishery goematics
- R Li
- H Li
- W Zou
- R G Smith
- T A Curran
Li, R., Li, H., Zou, W., Smith, R.G., Curran, T.A., 1996. An underwater digital photogrammetric system for fishery goematics. Int. Arch. Photogr. Remote Sens. 31 (B5), 524–529.
A comparison of the accuracy and precision of digital and analogue stereo–video systems
- Harvey
An underwater digital photogrammetric system for fishery goematics
- Li