Content uploaded by Qingqing L.
Author content
All content in this area was uploaded by Qingqing L. on Nov 27, 2020
Content may be subject to copyright.
Towards Active Vision with UAVs in Marine Search and Rescue:
Analyzing Human Detection at Variable Altitudes
Li Qingqing1, Jussi Taipalmaa2, Jorge Pe ˜
na Queralta1, Tuan Nguyen Gia1, Moncef Gabbouj2,
Hannu Tenhunen1, Jenni Raitoharju3, Tomi Westerlund1
1Turku Intelligent Embedded and Robotic Systems Lab, University of Turku, Finland
Emails: 1{qingqli, jopequ, tunggi, hatenhu, tovewe}@utu.fi
2Department of Computing Sciences, Tampere University, Finland
Emails: 2{jussi.taipalmaa, moncef.gabbouj}@tuni.fi
3Programme for Environmental Information, Finnish Environment Institute, Jyv¨
askyl¨
a, Finland
Email: jenni.raitoharju@environment.fi
Abstract—Unmanned Aerial Vehicles (UAVs) have been play-
ing an increasingly active role in supporting search and rescue
(SAR) operations in recent years. The benefits are multiple
such as enhanced situational awareness, status assessment, or
mapping of the operational area through aerial imagery. Most of
these application scenarios require the UAVs to cover a certain
area. If the objective is to detect people or other objects, or
analyze in detail the area, then there is a trade-off between
speed (higher altitude coverage) and perception accuracy (lower
altitude). An optimal point in between requires active perception
on-board the UAV to dynamically adjust the flight altitude and
path planning. As an initial step towards active vision in UAV
search in maritime SAR scenarios, in this paper we focus on
analyzing how the flight altitude affects the accuracy of object
detection algorithms. In particular, we quantify what are the
probabilities for false negatives and false positives in human
detection at different altitudes. Our results define the correlation
between the altitude and the ability of UAVs to effectively detect
people in the water.
Index Terms—Active Vision; Flight Altitude; Dynamic Alti-
tude; Object Detection; Human Detection; Marine Search and
Rescue (SAR); Unmanned Aerial Vehicles (UAV);
I. INTRODUCTION
Recent years have seen an increasingly wider adoption of
unmanned aerial vehicles (UAVs) to support search and res-
cue (SAR) operations. Owing to their fast deployment, speed
and aerial point of view, UAVs can aid quick response teams,
but also in longer-term monitoring and surveillance [1]. Some
of the main applications of UAVs in these scenarios are real-
time mapping of the operational area or delivery of emer-
gency supplies. In particular, UAVs can bring a significant
increase of the response team’s situational awareness and
detect objects and people from the air, specially those in
need of rescue [2]. An overview of recent research in this
area is available in [3], where UAVs for SAR operations
are characterized based on the operational environment, the
type of robotic systems in use, and the onboard sensing
capabilities of the UAVs.
We are interested in optimizing the support that UAVs
can provide in maritime SAR operations (see Fig. 1), but
Fig. 1: Illustration of active-vision-based search in maritime envi-
ronments with UAVs. A single UAV can first fly higher to cover
larger areas and descend in the event of a positive detection to in-
crease reliability. Search time can then be optimized by dynamically
adjusting the altitude depending on the perception confidence.
also for monitoring and surveillance in maritime environ-
ments, where they have already been widely utilized [4].
Maritime SAR operations might occur in both normal and
harsh environments. For example, according to the Spanish
national drowning report [5], in 2019 over 40% of drownings
happened on a beach, around 60% of the incidents happened
between 10:00 and 18:00, and in 20% of the cases lifeguards
were present in the area. Therefore, there is still a need
for better solutions for monitoring and supporting SAR
operations in safeguarded beaches, lakes or rivers even with
favorable weather conditions, which can then be extended
towards rougher environments as the technology evolves.
In this paper, we study the detection of people in mostly
still waters at different altitudes. In the future, we aim to
utilize this information within an active vision algorithm
that can dynamically adapt the flight plan of UAVs towards
optimization of search speed and reliability.
In terms of UAV-based perception, deep learning (DL)
methods have become the de-facto standards in object de-
tection and image segmentation with great success across
multiple domains [6], [7]. In this paper, we utilize the
YOLOv3 [6] architecture and characterize its performance for
human detection on still water surfaces. Within the machine
perception field, active vision has been a topic of interest
that has gained increasing research interest, owing to the
multiple advances in DL and accessibility of UAV platforms
for research. Active vision has been successfully applied for
single and multi-agent tracking [8], but we have observed a
gap in the literature in terms of active vision for search and
area coverage. The most active research direction in active
perception is currently reinforcement learning (RL) [9]. How-
ever, we consider in this paper a more traditional approach.
An RL approach can be challenging owing to the lack of
realistic simulators to train models for sea SAR.
Deep learning for perception in maritime environments
is limited by the lack of realistic training datasets openly
available. Moreover, a key challenge for UAV-based person
search and detection in these environments is the relatively
small size of objects to be detected in comparatively large
areas to be searched [10]. There is an evident trade-off
between speed and area coverage, and reliability of both
positive and negative detection. An additional challenge is
that the view of people at sea from the air is only partial, as
a significant portion of the body is immersed in the water.
Water reflection and refraction effects might also distort the
shape. In order to train YOLOv3 to adapt to this scenario,
and owing to the lack of open data for detecting people in
water, we collected over 450 high-resolution images to train,
validate and test our model. The images have been taken at
altitudes ranging from 20 m to 120 m.
This is, to the best of our knowledge, the first paper to
analyze the perception accuracy for UAVs with RGB cameras
in maritime environments as a function of their altitude. The
results can be generalized by accounting for the size in pixels
of the persons to be detected assuming well-focused images.
Moreover, the retrained YOLO model outperforms the state-
of-the-art in object classification, as it has been trained to
detect people even when only their head emerges above the
water level. The retrained YOLO model can be applied for
people swimming but also standing near the shore in a beach.
The rest of the paper is organized as follows. Section
II briefly overviews previous research in active vision, on
one side, and maritime SAR operations supported by UAVs,
on the other. We then describe the main objectives of our
study in Section III, together with data acquisition and model
training details. Section IV reports our experimental results
and Section V concludes the work.
II. BACK GROU ND
Multiple works have demonstrated the benefits of integrat-
ing UAVs to maritime SAR operations [11], [12]. Typical
sensors onboard UAVs are RGB, RGB-D and thermal cam-
eras, 3D lidars, and inertial/positional sensors for GNSS and
altitude estimation [13], [14]. With these sensors, UAVs can
aid in SAR operations by mapping the environment, locating
victims and survivors, and recognising and classifying dif-
ferent objects [13]. From the perception point of view, DL
methods have become the predominant solution for detecting
humans or other objects [7], [15], [16].
Human detection is a sub-task of object detection that
is of particular interest for SAR robotics [17]. Some of
the most popular neural network architectures for object
detection are R-CNN [18], Fast-RCNN [19], and YOLO [6].
In particular, YOLOv3 is the current state-of-the-art for real-
time detection, able of fast inference and high accuracy [6].
In this paper, we re-train the YOLOv3 network with a new
dataset for detecting people in the water.
Active perception has been defined as:
An agent is an active perceiver if it knows
why it wishes to sense, and then chooses what to
perceive, and determines how,when, and where to
achieve that perception. [20]
In UAV-aided maritime SAR operations, algorithms for area
coverage and human search incorporating active vision need
to be aware that their main objective is to find humans (why),
and need to be able to dynamically adjust their path planning
and orientation to achieve higher-confidence results (what).
This latter aspect can be achieved by, for instance, adjusting
their height and camera pitch, or by moving around the
person to get a better angle (how, where and when).
Active vision has been increasingly adopted in different
object detection tasks. However, no previous research has,
to the best of our knowledge, focused on active vision for
detection of humans in SAR scenarios. We therefore list
here some other relevant works in the area. Ammirato et
al. presented a dataset for robotic vision tasks in indoor
environments using RGBD cameras with the introduction
of an active vision strategy using Deep RL to predict the
next best move for object detection [21]. Juan et al. pre-
sented an autonomous Sequential Decision Process (SDP)
for active perception of targets in uncertain and cluttered
environments, with experiments conducted in a simulated
SAR scenario [22]. Davide et al. applied active vision to a
path planing algorithms that enabled quadrotor flight through
narrow gaps in indoor complex environments [23]. Manuela
et al. applied bio-inspired active vision for object avoidance
with wheel robots in indoor environments [24]. In SAR
operations, once a target has been identified, continuously
updating the position of target is essential, so that path
planning for the rescue teams can be adjusted. This can be
achieved though active tracking [25].
In terms of detecting people in maritime environments,
Eleftherios et al. presented a real-time human detection
system using DL models that run on-board UAVs to detect
open water swimmers [26]. The authors, however, do not
study the accuracy of the perception for different altitudes
or positions. In this work, we focus on analyzing human
detection as a trade-off between larger area coverage (higher
altitude) and higher amount of detail in the images (lower
altitudes).
(a) Beach view (b) Top view
(c) Low altitude (d) Back light
Fig. 2: (a) Example images of terrain at Littoinen Lake, Finland,
(b) The top view of swimmer, (c) The far view of swimmers, (d)
The close view of swimmer
In general, we see a clear trend towards a more widespread
utilization of UAVs in SAR missions and DL models for per-
ception (either onboard or offloading computation). We have
found, however, no previous works exploring the correlation
between the altitude at which UAVs fly and the detection
accuracy in maritime SAR scenarios.
III. METHODOLOGY
This section describes the data and details of the training
process for the perception algorithm. We also outline the
metrics that are analyzed in our experiments.
A. Data Acquisition
Owing to the lack of labeled open data showing people in
water, and in particular data labeled with the flight altitude,
we have collected data from people swimming and standing
in a lake. The dataset contains 458 labeled photos that are
taken by the camera mounted on the UAV. The camera has
a fixed focal length of 24 mm (35 mm format equivalent)
with a field of view of 83°and an aperture f/2.8. The images
have a resolution of 9 MP (4000 by 2250 pixels), and were
recorded near the beach area of Littoistenj¨
arvi Lake (60.4582
N, 22.3766 E), shown in Fig. 2 (a), Turku, Finland.
Each photo captures one or more people that are either
swimming or standing in the lake at different heights and
angles. Some examples are shown in Fig. 2 (b), (c) and (d).
However, the majority of pictures were taken with a gimbal
pitch of -90°(top-view images). The dataset contains 2D
bounding boxes for two classes: persons and other objects,
the latter one being used for animals in the water and other
floating objects. In addition to the bounding boxes, each
image contains information about the GPS position, relative
altitude to the take-off point (just above the water level), and
pitch angle of the camera gimbal (from horizontal images
with 0°pitch to top-view images with -90°pitch). The relative
altitude ranges from 0m to 143 m. While the dataset has
been acquired with good weather conditions and mostly still
waters, variable light conditions are also introduced. This
results in different colors for both water and people, as can
be seen in Fig. 2 (b) and (d). Some of the swimmers use
swimming caps of different colors and wear different types
of swimming suits.
B. Training and test setup
Training and testing were done with the YOLOv3 real-time
object detection model [6]. The YOLOv3 model pre-trained
with ImageNet [27] was trained again with our dataset using
transfer learning. Training is done in a way where all but
the last three layers are frozen for the first 50 epochs and
then unfrozen and trained further for another 50 epochs with
batch size of 32 and learning rate of 0.001.
Each image contains between 1 and 50 object instances.
The objects are divided into two classes: ’person’, contain-
ing 2454 instances, and ’something else’, containing 238
instances, mostly birds but also some other objects floating
in the water. All the images were labeled manually, using
bounding boxes with the Labelbox annotation tool [28].
Training and testing were done using 4-fold cross-validation,
randomly splitting the images using a 75/25 train/test split.
We refer to the re-trained model as the task-specific model
hereinafter.
C. Metrics
Object detection performance was evaluated using PAS-
CAL VOC challenge metrics [29] provided by [30]. We
calculated average perception (AP) for both classes separately
and mean average perception (mAP) over both classes using
different intersection over union (IoU) thresholds. The com-
parison in performance was done between the pre-trained
YOLOv3 model and the task-specific model with our data
using transfer learning. Furthermore, since our objective is
to analyze the correlation between the performance of the
human detection and the altitude, we also analyze how the
detection confidence and the ratio of false positives and false
negatives changes as a function of the altitude.
IV. EXP ERIMENTAL RES ULTS
In this section, we assess the performance of the trained
model as a classifier using the mean average precision for
different IoU thresholds, but also its usability for active-
vision-based control where the input to the algorithm is the
confidence of the model on each of its detections.
Some representative example detections made by the task-
specific model are illustrated in Fig. 3. In Fig. 3a, we observe
how the network is able to pinpoint the location of people in
the image, but the bounding box appears around the turbulent
water rather than around the person itself. However, not all
objects or turbulent areas are detected as people, as other
objects are also properly identified (Fig. 3b. In Fig. 3b, we
also observe that people can be located far away when the
gimbal pitch is closer to 0°. Finally, we see that even at high
altitudes, the confidence remains high and people are detected
also when immersed (Fig. 3c).
The performance of the task-specific model compared to
the pre-trained YOLOv3 network is shown in Table I, where
we see that the task-specific model is clearly superior. In
TABLE I: mAP-scores for different IoU-thresholds.
IoU-threshold
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Model
Task-specific model 0.6985 0.6984 0.6972 0.6954 0.6934 0.6883 0.6780 0.6384 0.5422 0.0000
Pre-trained YOLOv3 0.0547 0.0533 0.0533 0.0528 0.0514 0.0514 0.0514 0.0507 0.0440 0.0000
(a) Detection of one person (high confidence), and turbulent water
next to another (lower confidence). Altitude: 37 m. Pitch: -80°.
(b) Detection of other objects but missing two persons in the
distance. Altitude: 12 m. Pitch: -25°.
(c) Successful detection of three people at high altitude, one of them fully immersed in the water (only a portion of the original image
is shown). Altitude: 86 m. Gimbal pitch: -90°.
Fig. 3: Samples of detections made using the task-specific model.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0.8
0.84
0.88
0.92
0.96
1
Recall
Precision
Person
Others
Fig. 4: Precision x Recall curve for class ’person’ and ’some-
thing˙else’ using IoU-threshold 0.5 with the task-specific model.
terms of the precision ×recall curves, those corresponding to
classes ’person’ and ’something else’ are provided in Fig. 4.
Next, we analyze performance at different altitudes. The
significance of the altitude is, however, relative to the reso-
lution of the camera and its ability to produce clear images.
20 40 60 80 100 120 140
0
50
100
150
200
250
Relative Altitude (m)
Bounding box side length (pixels)
Individual
Average
Fig. 5: Side length of the ground truth bouding boxes, in pixels,
based on the altitude.
The camera pitch is also important as illustrated. In order to
provide results that are more generalizable, Fig. 5 shows the
size in pixels of the ground truth bounding boxes.
Fig. 6a shows all the person detections plotted in terms
of their confidence against the altitude, using IoU = 0.1to
20 40 60 80 100 120
0.2
0.4
0.6
0.8
1
Altitude (m)
Detection Confidence
True Positives
False Positives
(a) Confidence with IOU
20 40 60 80 100 120
0.2
0.4
0.6
0.8
1
Altitude (m)
Detection Confidence
True Positives
False Positives
(b) Confidence with DIST
Fig. 6: Confidence of individual detections as a function of the relative UAV altitude. We observe a clear difference between high-confidence
and true positives under the threshold of 100 m, with lower confidence and higher rate of false positives above it.
TP/IOU FP/IOU TP/DIST FP/DIST
0.2
0.4
0.6
0.8
1
Distributions
Confidence
Fig. 7: Distributions for the confidence of true positive (TP) and
false positive (FP) detections (DIST and IoU = 0.1).
20 40 60 80 100 120
0
0.25
0.5
0.75
1
Relative altitude (m)
FN/(TP+FN) Proportion
IoU=0.1
IoU=0.5
DIST
Fig. 8: Proportion of false negatives (FN) over true positives (TP)
and FN. This gives an idea of the probability of missing a person.
consider true positives. We have set the IoU to 0.1 because
we are only interested in pointing to the approximate location
of persons but not their exact size and place. For altitudes
under 60 m, over 98.8% of the detections with a confidence
above 0.5 are correct. A clear threshold appears at an altitude
of 90 m. Above 90 m, 83.3% of the detections are correct.
In some of the test images, we have noticed that the model
detects turbulence in the water created by people as persons,
and not the full bodies of the people themselves. Because we
are not interested in analyzing how capable the task-specific
model is of generating accurate bounding boxes, but instead
on pointing to the approximate location of people at sea,
we might also want to consider as correct detection boxes
that are just adjacent to actual people. In Fig. 6b, we have
plotted the confidence as a function of the altitude, but now
using a distance in pixels of less than 100 between the ground
truth and the predicted box (DIST) to assume that a detection
is correct. We now see that all except one of the positive
detections with a confidence of over 70% are correct for an
altitude up to 100 m. For a confidence above 45%, all but
one detections are correct up to an altitude of 70 m. The
distributions of the true positives and false positives for each
of the two metrics (IoU, DIST) are shown in Fig. 7. There is
a clear threshold just under a confidence of 0.6, with almost
75% of true positive having a confidence over 0.6, and almost
75% of false positives having a lower confidence.
In order to evaluate this model within its context for
SAR missions, we also need to take into account that false
positives do not necessarily have a significant impact on the
search performance, but false negatives do, as they mean
that the UAV misses a person. We have therefore plotted in
Fig. 8 the proportion of false negatives over true positives. If
we use the pixel distance to consider a detection as correct,
then the proportion remains under 10% for all altitudes. With
IoU = 0.5, however, over 50% of the people in the water
are undetected. However, we do not consider this an effective
way of evaluating a detection in this scenario.
V. CONCLUSION
With UAVs increasingly penetrating multiple civil domains
and, among them, search and rescue operations, more com-
plex control mechanisms are required for more autonomous
UAVs. To that end, active perception is one of the most
promising research directions. In UAV search, active vision
can be exploited to optimize the flight plan based on the
confidence of the DL vision algorithms. We have presented
preliminary work that studies the confidence of a re-trained
YOLOv3 model for detecting people in the water for altitudes
ranging from 20 m to 120 m. With a custom dataset, we
have seen a major performance increase with respect to
the pre-trained YOLOv3 network. Our results show a clear
correlation between the altitude and the confidence of the
detections and between the confidence and the correctness of
the detections. When considering as true positives detections
near actual people (e.g., over water turbulence created by
people), we have seen that the proportion of false negatives
remains low even for high altitudes, and the proportion of
false positives over true positives drops significantly for all
predictions with a confidence over 60%. Finally, we have
observed a clear altitude threshold at around 100 m after
which confidence and accuracy drop.
The results presented in this paper will serve as the starting
point towards the design of active-vision-based search with
UAVs in marine SAR operations. In future works, we will
also incorporate the camera pitch into the analysis. The
dataset will be made publicly available with further additions.
ACK NOWLEDGEM ENT S
This work was supported by the Academy of Finland’s
AutoSOS project with grant number 328755.
REFERENCES
[1] H. Shakhatreh, A. H. Sawalmeh, A. Al-Fuqaha, Z. Dou, E. Almaita,
I. Khalil, N. S. Othman, A. Khreishah, and M. Guizani, “Unmanned
aerial vehicles (uavs): A survey on civil applications and key research
challenges,” IEEE Access, vol. 7, pp. 48572–48 634, 2019.
[2] J. Pe˜
na Queralta, J. Taipalmaa, B. C. Pullinen, V. K. Sarker, T. N. Gia,
H. Tenhunen, M. Gabbouj, J. Raitoharju, and T. Westerlund, “Collab-
orative multi-robot search and rescue: Coordination and perception,”
arXiv preprint arXiv:2008.12610 [cs.RO], 2020.
[3] S. Grogan, R. Pellerin, and M. Gamache, “The use of unmanned
aerial vehicles and drones in search and rescue operations–a survey,”
Proceedings of the PROLOG, 2018.
[4] W. Roberts, K. Griendling, A. Gray, and D. Mavris, “Unmanned
vehicle collaboration research environment for maritime search and
rescue,” in 30th Congress of the International Council of the Aero-
nautical Sciences. International Council of the Aeronautical Sciences
(ICAS) Bonn, Germany, 2016.
[5] Royal Spanish Federation of First Aid and Rescue, “National Drown-
ings Report - Informe Nacional de Ahogamientos (INA),” 2019.
[6] J. Redmon and A. Farhadi, “YOLOv3: An incremental improvement,”
arXiv, 2018.
[7] S.-J. Hong, Y. Han, S.-Y. Kim, A.-Y. Lee, and G. Kim, “Application of
deep-learning methods to bird detection using unmanned aerial vehicle
imagery,” Sensors, vol. 19, no. 7, p. 1651, 2019.
[8] R. Tallamraju, E. Price, R. Ludwig, K. Karlapalem, H. H. B¨
ulthoff,
M. J. Black, and A. Ahmad, “Active perception based formation control
for multiple aerial vehicles,” IEEE Robotics and Automation Letters,
vol. 4, no. 4, pp. 4491–4498, 2019.
[9] D. Gallos and F. Ferrie, “Active vision in the era of convolutional
neural networks,” in 2019 16th Conference on Computer and Robot
Vision (CRV), 2019, pp. 81–88.
[10] J. Pe˜
na Queralta, J. Raitoharju, T. N. Gia, N. Passalis, and T. Wester-
lund, “Autosos: Towards multi-uav systems supporting maritime search
and rescue with lightweight ai and edge computing,” arXiv preprint
arXiv:2005.03409, 2020.
[11] A. Matos, A. Martins, A. Dias, B. Ferreira, J. M. Almeida, H. Ferreira,
G. Amaral, A. Figueiredo, R. Almeida, and F. Silva, “Multiple robot
operations for maritime search and rescue in eurathlon 2015 competi-
tion,” in OCEANS 2016-Shanghai. IEEE, 2016, pp. 1–7.
[12] J. G¨
uldenring, L. Koring, P. Gorczak, and C. Wietfeld, “Heterogeneous
multilink aggregation for reliable uav communication in maritime
search and rescue missions,” in 2019 International Conference on
Wireless and Mobile Computing, Networking and Communications
(WiMob). IEEE, 2019, pp. 215–220.
[13] R. Konrad, D. Serrano, and P. Strupler, “Unmanned aerial systems,”
Search and Rescue Robotics—From Theory to Practice, pp. 37–52,
2017.
[14] H. Surmann, R. Worst, T. Buschmann, A. Leinweber, A. Schmitz,
G. Senkowski, and N. Goddemeier, “Integration of uavs in urban
search and rescue missions,” in 2019 IEEE International Symposium
on Safety, Security, and Rescue Robotics (SSRR). IEEE, 2019, pp.
203–209.
[15] T. Giitsidis, E. G. Karakasis, A. Gasteratos, and G. C. Sirakoulis,
“Human and fire detection from high altitude uav images,” in 2015
23rd Euromicro International Conference on Parallel, Distributed, and
Network-Based Processing. IEEE, 2015, pp. 309–315.
[16] S. Yong and Y. Yeong, “Human object detection in forest with deep
learning based on drone’s vision,” in 2018 4th International Conference
on Computer and Information Sciences (ICCOINS). IEEE, 2018, pp.
1–5.
[17] “Autonomous human detection system mounted on a drone,” 2019
International Conference on Wireless Communications, Signal Pro-
cessing and Networking, WiSPNET 2019, pp. 335–338, 2019.
[18] R. Girshick, J. Donahue, T. Darrell, and J. Malik, “Rich feature
hierarchies for accurate object detection and semantic segmentation,”
in Proceedings of the IEEE conference on computer vision and pattern
recognition, 2014, pp. 580–587.
[19] R. Girshick, “Fast r-cnn,” in International Conference on Computer
Vision (ICCV), 2015.
[20] R. Bajcsy, Y. Aloimonos, and J. Tsotsos, “Revisiting active perception,”
Autonomous Robots, vol. 42, pp. 177—-196, 2018.
[21] P. Ammirato, P. Poirson, E. Park, J. Koˇ
seck´
a, and A. C. Berg, “A
dataset for developing and benchmarking active vision,” in 2017 IEEE
International Conference on Robotics and Automation (ICRA), 2017,
pp. 1378–1385.
[22] J. Sandino, F. Vanegas, F. Gonz´
alez, and F. Maire, “Autonomous uav
navigation for active perception of targets in uncertain and cluttered
environments,” in 2020 IEEE Aerospace Conference, 2020.
[23] D. Falanga, E. Mueggler, M. Faessler, and D. Scaramuzza, “Aggressive
quadrotor flight through narrow gaps with onboard sensing and com-
puting using active vision,” in 2017 IEEE International Conference on
Robotics and Automation (ICRA), 2017, pp. 5774–5781.
[24] M. Chessa, S. Murgia, L. Nardelli, S. P. Sabatini, and F. Solari,
“Bio-inspired active vision for obstacle avoidance,” in Conference on
Computer Graphics Theory and Applications, 2014, pp. 1–8.
[25] F. Zhong, P. Sun, W. Luo, T. Yan, and Y. Wang, “AD-VAT: An
asymmetric dueling mechanism for learning visual active tracking,”
in International Conference on Learning Representations, 2019.
[26] E. Lygouras, N. Santavas, A. Taitzoglou, K. Tarchanidis, A. Mitropou-
los, and A. Gasteratos, “Unsupervised human detection with an em-
bedded vision system on a fully autonomous uav for search and rescue
operations,” Sensors, vol. 19, no. 16, p. 3542, 2019.
[27] J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei,
“ImageNet: A Large-Scale Hierarchical Image Database,” in CVPR09,
2009.
[28] Labelbox, “Labelbox,” 2019, [Online]. Available: https://labelbox.com.
[29] M. Everingham, L. Van Gool, C. K. I. Williams, J. Winn, and
A. Zisserman, “The pascal visual object classes (voc) challenge,”
International Journal of Computer Vision, vol. 88, no. 2, pp. 303–338,
Jun. 2010.
[30] R. Padilla, S. L. Netto, and E. A. B. da Silva, “A survey on perfor-
mance metrics for object-detection algorithms,” in 2020 International
Conference on Systems, Signals and Image Processing (IWSSIP), 2020,
pp. 237–242.
