No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Benchmarking YOLOv8 for Optimal Crack Detection in Civil Infrastructure
Abstract
Ensuring the structural integrity and safety of bridges is crucial for the reliability of transportation networks and public safety. Traditional crack detection methods are increasingly being supplemented or replaced by advanced artificial intelligence (AI) techniques. However, most of the models rely on two-stage target detection algorithms, which pose concerns for real-time applications due to their lower speed. While models such as YOLO (You Only Look Once) have emerged as transformative tools due to their remarkable speed and accuracy. However, the potential of the latest YOLOv8 framework in this domain remains underexplored. This study bridges that gap by rigorously evaluating YOLOv8's performance across five model scales (nano, small, medium, large, and extra-large) using a high-quality Roboflow dataset. A comprehensive hyperparameter optimization was performed, testing six state-of-the-art optimizers— Stochastic Gradient Descent, Adaptive Moment Estimation, Adam with Decoupled Weight Decay, Root Mean Square Propagation, Rectified Adam, and Nesterov-accelerated Adam. Results revealed that YOLOv8, optimized with Stochastic Gradient Descent, delivered exceptional accuracy and speed, setting a new benchmark for real-time crack detection. Beyond its immediate application, this research positions YOLOv8 as a foundational approach for integrating advanced computer vision techniques into infrastructure monitoring. By enabling more reliable and proactive maintenance of aging bridge networks, this work paves the way for safer, more efficient transportation systems worldwide.
ResearchGate has not been able to resolve any citations for this publication.
In this study, we propose a YOLOv8-based method for detecting bird nests on transmission lines, aiming to solve the safety hazards caused by transmission lines being affected by nesting birds. First, we collect a large-scale image dataset containing transmission lines and label the bird nests in it. Then, we used YOLOv8 as a target detection model to enable accurate detection of bird nests in transmission lines by end-to-end training on the dataset. During the training process, we adopt Bounding Box Loss, Vision Feature Loss, and Classification Loss loss functions to help the model learn more accurate and meaningful feature representations and improve the detection performance. The experimental results show that our proposed method achieves high accuracy and recall on the bird nest detection task, which can effectively help the transmission line maintenance personnel to detect and deal with bird nests in time, and ensure the safe and stable operation of transmission lines.
Surface cracks are alluded to as one of the early signs of potential damage to infrastructures. In the same vein, their detection is an imperative task to preserve the structural health and safety of bridges. Human-based visual inspection is acknowledged as the most prevalent means of assessing infrastructures’ performance conditions. Nonetheless, it is unreliable, tedious, hazardous, and labor-intensive. This state of affairs calls for the development of a novel YOLOv8-AFPN-MPD-IoU model for instance segmentation and quantification of bridge surface cracks. Firstly, YOLOv8s-Seg is selected as the backbone network to carry out instance segmentation. In addition, an asymptotic feature pyramid network (AFPN) is incorporated to ameliorate feature fusion and overall performance. Thirdly, the minimum point distance (MPD) is introduced as a loss function as a way to better explore the geometric features of surface cracks. Finally, the middle aisle transformation is amalgamated with Euclidean distance to compute the length and width of segmented cracks. Analytical comparisons reveal that this developed deep learning network surpasses several contemporary models, including YOLOv8n, YOLOv8s, YOLOv8m, YOLOv8l, and Mask-RCNN. The YOLOv8s + AFPN + MPDIoU model attains a precision rate of 90.7%, a recall of 70.4%, an F1-score of 79.27%, mAP50 of 75.3%, and mAP75 of 74.80%. In contrast to alternative models, our proposed approach exhibits enhancements across performance metrics, with the F1-score, mAP50, and mAP75 increasing by a minimum of 0.46%, 1.3%, and 1.4%, respectively. The margin of error in the measurement model calculations is maintained at or below 5%. Therefore, the developed model can serve as a useful tool for the accurate characterization and quantification of different types of bridge surface cracks.
The accuracy of detecting superficial bridge defects using the deep neural network approach decreases significantly under light variation and weak texture conditions. To address these issues, an enhanced intelligent detection method based on the YOLOv8 deep neural network is proposed in this study. Firstly, multi-branch coordinate attention (MBCA) is proposed to improve the accuracy of coordinate positioning by introducing a global perception module in coordinate attention mechanism. Furthermore, a deformable convolution based on MBCA is developed to improve the adaptability for complex feature shapes. Lastly, the deformable convolutional network attention YOLO (DCNA-YOLO) detection algorithm is formed by replacing the deep C2F structure in the YOLOv8 architecture with a deformable convolution. A supervised dataset consisting of 4794 bridge surface damage images is employed to verify the proposed method, and the results show that it achieves improvements of 2.0% and 3.4% in mAP and R. Meanwhile, the model complexity decreases by 1.2G, increasing the detection speed by 3.5/f·s⁻¹.
Road crack detection is one of the important parts of road safety detection. Aiming at the problems such as weak segmentation effect of basic U-Net on pavement crack, insufficient precision of crack contour segmentation, difficult to identify narrow crack and low segmentation accuracy, this paper proposes an improved U-net network pavement crack segmentation method. VGG16 and Up_Conv (Upsampling Convolution) modules are introduced as backbone network and feature enhancement network respectively, and the more abstract features in the image are extracted by using the Block depth separable convolution blocks, and the multi-scale features are captured and enhanced by higher level semantic information to improve the recognition accuracy of narrow cracks in the road surface. The improved network embedded the Ca(Channel Attention) attention mechanism in U-net’s jump connection to enhance the crack portion to suppress background noise. At the same time, DG_Conv(Depthwise GSConv Convolution) module and UnetUp(Unet Upsampling) module are added in the decoding part to extract richer features through more convolutional layers in the network, so that the model pays more attention to the detailed part of the crack, so the segmentation accuracy can be improved. In order to verify the model’s ability to detect cracks in complex backgrounds, experiments were carried out on CFD and Deepcrack datasets. The experimental results show that compared with the traditional U-net network F1-score and mIoU have increased by 13.6% and 9.9% respectively. Superior to advanced models such as U-net, Segnet and Linknet in accuracy and generalization ability, the improved model provides a new method for asphalt pavement crack detection. The model is more conducive to practical application and ground deployment, and can be applied in road maintenance projects.
A bridge disease identification approach based on an enhanced YOLO v3 algorithm is suggested to increase the accuracy of apparent disease detection of concrete bridges under complex backgrounds. First, the YOLO v3 network structure is enhanced to better accommodate the dense distribution and large variation of disease scale characteristics, and the detection layer incorporates the squeeze and excitation (SE) networks attention mechanism module and spatial pyramid pooling module to strengthen the semantic feature extraction ability. Secondly, CIoU with better localization ability is selected as the loss function for training. Finally, the K-means algorithm is used for anchor frame clustering on the bridge surface disease defects dataset. 1363 datasets containing exposed reinforcement, spalling, and water erosion damage of bridges are produced, and network training is done after manual labelling and data improvement in order to test the efficacy of the algorithm described in this paper. According to the trial results, the YOLO v3 model has enhanced more than the original model in terms of precision rate, recall rate, Average Precision (AP), and other indicators. Its overall mean Average Precision (mAP) value has also grown by 5.5%. With the RTX2080Ti graphics card, the detection frame rate increases to 84 Frames Per Second, enabling more precise and real-time bridge illness detection.
The combination of UAV camera and intelligent algorithm is a promising method for non-contact bridge crack detection. In this paper, an inspection tool based on UAV Image Acquisition Technology (UAVIAT) and Improved Intelligent Target Detection Technology (IITDT) called Improved Intelligent Real-Time Crack Detection Method for Bridges (IIRTCDMB) is proposed for efficient crack detection. The contributions of this paper are (1) The Squeeze-Excitement (SE) attention module is integrated into the target detection algorithm - You Only Look Once version 7 (YOLOv7) model to improve the learning ability of the feature channel. A Focal-efficient intersection over union (Focal-EIoU) loss function is also introduced to improve the regression accuracy of the model. As a result, a new crack image detection algorithm, YOLOv7-CD, is proposed. (2) A training process based on two-stage transfer learning (TSTL) is established, and hyper-parameter optimization of YOLOv7-CD is carried out. The feasibility and excellent performance of the proposed method are verified by applying it on the Cuntan Yangtze River Bridge. The results show that the average precision (AP) of the YOLOv7-CD model is improved by 3.19% compared with the original YOLOv7 model. After TSTL and hyperparameter optimization, the AP of the YOLOv7-CD model for bridge crack detection reaches 98.01%, which is higher than that of the popular target detection models. The IIRTCDMB proposed in this paper can acquire bridge surface images more safely and efficiently, and provide inspectors with more accurate structural crack information with lower computational and hardware requirements, which can provide technical support for the assessment of structural safety conditions and the formulation of maintenance programs.
The maintenance of pavements takes considerable time and poses a significant task, especially when it comes to detecting cracks at the pixel level. Due to the complexity of pavement conditions, such as road markings, shadows, and oil stains, deep learning techniques are still a challenge in automating crack detection. This paper presents a novel methodology termed as CrackHAM, which is an encoder-decoder network founded on the U-Net architecture. The primary objectives of CrackHAM are twofold: to achieve accurate and robust pavement crack detection while reducing the parameters of the network. Our study introduces two significant improvements to the existing neural network architecture, namely the phased multi-fusion module and the dual attention mechanisms. These improvements improve the process of defect extraction, resulting in an improved level of performance. Furthermore, a novel module named HASPP is devised to augment the network’s capacity to acquire more comprehensive receptive fields. In order to lower the number of network parameters, a technique is employed whereby only use half of the number of input channels and output channels in the VGG16 are utilized as U-Net encoder modules. The empirical findings demonstrate that in the Deepcrack, Crack500, and FIND public datasets, CrackHAM achieves superior segmentation performance compared to the FCN, Deeplabv3, Swin-Unet, and U-Net models while utilizing only one-third of the computational resources.
Owing to the development of computerized vision technology, object detection based on convolutional neural networks is being widely used in the field of bridge crack detection. However, these networks have limited utility in bridge crack detection because of low precision and poor real-time performance. In this study, an improved single-shot multi-box detector (SSD) called ISSD is proposed, which seamlessly combines the depth separable deformation convolution module (DSDCM), inception module (IM), and feature recalibration module (FRM) in a tightly coupled manner to tackle the challenges of bridge crack detection. Specifically, DSDCM was utilized for extracting the characteristic information of irregularly shaped bridge cracks. IM was designed to expand the width of the network, reduce network calculations, and improve network computing speed. The FRM was employed to determine the importance of each feature channel through learning, enhance the useful features according to their importance, and suppress the features that are insignificant for bridge crack detection. The experimental results demonstrated that ISSD is effective in bridge crack detection tasks and offers competitive performance compared to state-of-the-art networks.
Deep learning-based bridge crack detection methods have advantages over traditional methods. We proposed an automated bridge crack detection method using lightweight vision models. First, our study applied the You Only Look Once 4th version (YOLO v4) (Bochkovskiy et al. in Yolov4: Optimal speed and accuracy of object detection. arXiv:200410934, 2020) to bridge surface crack detection. Then, to achieve model acceleration, some lightweight networks were used to replace the feature extraction network in YOLO v4, which reduced the parameter numbers and the backbone layers. The lightweight design can reduce the computational overhead of the model, making it convenient to deploy on edge platforms with limited computational power. The experimental results showed that the lightweight network-based bridge crack detection model required significantly less storage space at the expense of a slight reduction in precision. Therefore, an improved YOLO v4 crack detection method was proposed to meet real-time running without sacrificing accuracy. The precision, recall, and F1 score of the proposed crack detection method are 93.96%, 90.12%, and 92%, respectively. And the model only required 23.4 MB of storage space, and its frames per second could reach 140.2 frames. Compared with existing bridge crack detection methods, the proposed method showed precision, speed, and model size advantages.
The resistance of concrete to chloride penetration determines the durability and service life of reinforced concrete building structures in coastal or chloride-laden environments. This work adopted five machine learning algorithms, naïve bayes, k-nearest neighbors, decision trees, support vector machine, and random forests, to predict the chloride resistance level of concrete based on its ingredients, considering two scenarios. The first scenario considers all features describing the mix components, whereas the second scenario considers only a subset of the features. All models are validated by performing intensive evaluation matrices using unseen data. The validation results confirm that the developed models predict the level of chloride resistance of concrete with high accuracy. Of all the algorithms, the support vector machine performed best, with 89% and 88% accuracy in the first and second scenarios, respectively.
Crack detection of the concrete bridge is an essential index for the safety assessment of bridge structure. It is more important to check the whole structure than to check the accuracy in the damage assessment. However, the traditional deep learning model method cannot completely detect the crack structure, which challenges image‐based crack detection. For this reason, we propose deep bridge crack classification (DBCC)‐Net as a classification‐based deep learning network. By pruning the Yolox, the regression problem of the target detection is converted to the binary classification problem to avoid the network performance degradation caused by the translation invariance of the convolutional neural network (CNN). In addition, the network post‐processing and a two‐stage crack detection strategy are proposed to enable the network to detect cracks and extract crack morphology in high‐resolution images quickly. In the first stage, DBCC‐Net realizes the coarse extraction of crack position based on image slice classification. In the second stage, the complete crack morphology is extracted from the location suggested by the semantic segmentation network. Experimental results show that the proposed two‐stage method has 19 frames per second (FPS) and 0.79 Miou (mean intersection over union) at the actual bridge images with 2560×2560 pixels. Although FPS is reduced, the Miou value is 7.8% higher than other methods, proving this paper's practical value.
This study explored the performance of ten pre-trained CNN architectures in detecting and classifying asphalt
pavement cracks from images. A comparison of eight optimisation techniques led to developing an optimised pretrained CNN model tailored for crack classification, with DenseNet201 emerging as the most effective, closely
followed by ShuffleNet and ResNet101. Conversely, VGG16 exhibited notably lower accuracy among the models
evaluated. Through the application of diverse feature selection techniques as optimisers, DenseNet201 consistently outperformed others, followed by DarkNet19 and Xception. Despite employing different optimisers,
VGG16 and VGG19 consistently demonstrated inferior performance. The research introduced a novel approach
utilising the DenseNet201 model and the GWO optimiser for asphalt pavement crack classification, validated
against various CNN models. Its robustness was verified by testing against images contaminated with differing
levels and types of noise, yielding promising outcomes. Results underscore the method's potential for accurately
detecting diverse crack types, implying applicability in real-world scenarios.
Hong Kong, among the world’s most densely populated cities, has witnessed rapid growth in traffic volume, resulting in increased traffic density and vehicle loads. Regular bridge inspections are imperative to ensure human safety and safeguard property. However, conventional visual inspection methods are highly criticized for their critical limitations such as inaccuracy, subjectivity, labor-intensiveness, tediousness, and hazardousness. Cracks are regarded as the most prevalent type of defects encountered during inspection of reinforced concrete bridges. Automated detection of bridge surface cracks is a quite challenging and hectic task due to their random characteristics and usual in complex and non-uniform background textures. Presence. In light of foregoing, this paper proposes a novel computer vision model for concrete bridge crack detection in an attempt to circumvent the critical deficiencies of manual visual inspection. The developed model is envisioned on the use of you only look once version 8 (YOLOv8) architecture, which is cited as one of the most advanced convolutional neural networks structures for multi-scale object detection. Comprising three fundamental components - the backbone, neck, and head, this model introduces the concept of a decoupled head, segregating it into a detection head and a classification head. This design empowers the model with greater flexibility in handling diverse tasks. Moreover, the incorporation of the global attention module (GAM) and the wise intersection over union (IoU) loss function
serves to further boost detection correctness of the developed model and amplify its generalization ability. The developed YOLOv8-GAM-Wise-IoU is compared against some of the widely acknowledged one-stage and twostage deep learning models using the evaluation metrics of precision, recall, F1-score, mean average precision (mAP) and IoU. It outperformed them accomplishing testing precision, recall, F1-score, mAP50, mAP50–95 and mAP75 of 97.4%, 94.9%, 0.96, 98.1%, 76.2%, and 97.8%, respectively. It is also observed that developed model maintains a modest size of 93.20 M resulting in diminishing the computational cost of training and inference processes. This makes it highly deployable in various crack detection pertaining applications. It can be argued that the developed model can contribute notably to the preservation of safety and integrity of reinforced concrete
bridges in Hong Kong environment.
Detecting and measuring cracks on a bridge deck is crucial for preventing further damage and ensuring safety. However, manual methods are slow and subjective, highlighting the need for an efficient solution to detect and measure crack length and width. This study proposes a novel process-based deep learning approach for detecting and segmenting cracks on the bridge deck. Five state-of-the-art object detection networks were evaluated for their performance in detecting cracks: Faster RCNN-ResNet50, Faster RCNN-ResNet101, RetinaNet-ResNet50, RetinaNet-ResNet101, and YOLOv7. Additionally, two object segmentation networks, U-Net, and pix2pix, were optimized by experimenting with various network depths, activation functions, loss functions, and data augmentation to segment the detected cracks. The results showed that YOLOv7 outperformed both Faster RCNN and RetinaNet with both ResNet50 and ResNet101 backbones in terms of both speed and accuracy. Furthermore, the proposed U-Net is better than the mainstream U-Net and pix2pix networks. Based on these results, YOLOv7 and the proposed U-Net are integrated for detecting and segmenting cracks on a bridge deck. The proposed method was then applied to two bridges in South Korea to test its performance, and the results showed that it could detect crack length with an accuracy of 92.38 percent. Moreover, the proposed method can determine crack width and classify it with an R2 value of 0.87 and an average accuracy of 91 percent, respectively. In summary, this study provides an efficient and reliable method for detecting, measuring, and classifying cracks on a bridge deck surface.
You can download the paper by clicking to this link: https://authors.elsevier.com/a/1hXoj3O1E1UU6G
Recently, deep learning (DL) methods have been intensively studied in pavement distress detection research. To build an effective yet robust DL-based distress detection model, both the training data and DL model structure need to be carefully prepared. This paper discusses the main influencing factors that affect the performance of DL models in pavement distress detection and aims to explore possible ways to improve the model performance. An open-source pavement image dataset and three promising object detection methods were selected for model development and evaluation. A series of experiments were conducted with different settings of model structures and data manipulation. According to the experiment results, selection of detection methods, data size, and annotation consistency have significant impact on detection performance. DL models YOLO and Faster R-CNN were found to yield the same level of performance, while CenterNet notably underperformed on the test dataset. Distress classes with a larger number of instances were found to be able to yield better performances. With the new annotation data carefully marked up and verified by experienced pavement raters, the accuracy score using the F1 index was found to be 0.84, which was a very sound improvement compared with the 0.7 of the original annotation. Data augmentation methods were found to be able to significantly improve the detection accuracy. Based on the experiment results, we believe the focus should be placed on the development of high-quality training datasets and thin object detection methods in the future.
In order to improve the efficiency and accuracy of bridge crack detection, this paper introduces Deep Learning for bridge crack detection with Unmanned Aerial Vehicles (UAVs). In this study, a method for deciding the distance range for stable imaging was used, and we have done a set of verification experiments on three types of UAVs. Then a Two-RTSs system for UAV hovering accuracy was proposed, and the displacement and attitude angle changes of UAV were acquired. The results indicated that DJI M210-RTK could acquire high-quality images at longer distances and had better stability in hover mode. So, we decided to use the DJI M210-RTK for bridge crack recognition. Finally, the Faster Region Convolutional Neural Network (Fatesr R-CNN) algorithm based on VGG16 transfer learning is used for the apparent detection of a bridge located in Hunan Province. Our method can achieve precision, recall and F1-score of 92.03 %, 96.26 %, and 94.10 %, respectively. The outcome of this study shows that automatic bridge crack detection using UAVs and Faster R-CNN can be more efficient while maintaining high accuracy.
The appearance of surface cracks is an important feature of the early safety of the architecture being threatened, which is of great significance to maintaining the safety of the bridge. Bridge safety maintenance is still a challenging issue owing to noise interference and unclear bridge images. The detection method based on deep learning has achieved remarkable result in semantic segmentation, target detection, and other fields, but still be unable to focus on balancing speed and accuracy. Wherefore, we designed a new crack detection network‐dense boundary refinement network (DBR‐Net), which combines the advantages of STDC‐Net (Short‐Term Dense Concatenate network) and refinement network. STDC‐Net mainly improves the detection rate by eliminating redundant structures and optimizes the detailed information by binary cross‐entropy loss and dice loss. Finally, the detail aggregation module combines the shallow spatial details with the deep semantic information to predict the segmentation results. Among them, the refinement network refines the segmentation results of the offset map generated by the training stage and the test stage. We verified the feasibility of DBR‐Net on three datasets. At the same time, the detection accuracy rate on our self‐made crack dataset reached 97.54%. What's more worth mentioning is that our detection rate has reached 37.0 images per second (IPS), realizing real‐time crack detection.
We propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature. The new architecture leads to an automatically learned, unsupervised separation of high-level attributes (e.g., pose and identity when trained on human faces) and stochastic variation in the generated images (e.g., freckles, hair), and it enables intuitive, scale-specific control of the synthesis. The new generator improves the state-of-the-art in terms of traditional distribution quality metrics, leads to demonstrably better interpolation properties, and also better disentangles the latent factors of variation. To quantify interpolation quality and disentanglement, we propose two new, automated methods that are applicable to any generator architecture. Finally, we introduce a new, highly varied and high-quality dataset of human faces.
We present some updates to YOLO! We made a bunch of little design changes to make it better. We also trained this new network that's pretty swell. It's a little bigger than last time but more accurate. It's still fast though, don't worry. At 320x320 YOLOv3 runs in 22 ms at 28.2 mAP, as accurate as SSD but three times faster. When we look at the old .5 IOU mAP detection metric YOLOv3 is quite good. It achieves 57.9 mAP@50 in 51 ms on a Titan X, compared to 57.5 mAP@50 in 198 ms by RetinaNet, similar performance but 3.8x faster. As always, all the code is online at https://pjreddie.com/yolo/
The vanishing gradient problem was a major obstacle for the success of deep learning. In recent years it was gradually alleviated through multiple different techniques. However the problem was not really overcome in a fundamental way, since it is inherent to neural networks with activation functions based on dot products. In a series of papers, we are going to analyze alternative neural network structures which are not based on dot products. In this first paper, we revisit neural networks built up of layers based on distance measures and Gaussian activation functions. These kinds of networks were only sparsely used in the past since they are hard to train when using plain stochastic gradient descent methods. We show that by using Root Mean Square Propagation (RMSProp) it is possible to efficiently learn multi-layer neural networks. Furthermore we show that when appropriately initialized these kinds of neural networks suffer much less from the vanishing and exploding gradient problem than traditional neural networks even for deep networks.
We train generative 'up-convolutional' neural networks which are able to generate images of objects given object style, viewpoint, and color. We train the networks on rendered 3D models of chairs, tables, and cars. Our experiments show that the networks do not merely learn all images by heart, but rather find a meaningful representation of 3D models allowing them to assess the similarity of different models, interpolate between given views to generate the missing ones, extrapolate views, and invent new objects not present in the training set by recombining training instances, or even two different object classes. Moreover, we show that such generative networks can be used to find correspondences between different objects from the dataset, outperforming existing approaches on this task.
We present YOLO, a unified pipeline for object detection. Prior work on
object detection repurposes classifiers to perform detection. Instead, we frame
object detection as a regression problem to spatially separated bounding boxes
and associated class probabilities. A single neural network predicts bounding
boxes and class probabilities directly from full images in one evaluation.
Since the whole detection pipeline is a single network, it can be optimized
end-to-end directly on detection performance. Our unified architecture is also
extremely fast; YOLO processes images in real-time at 45 frames per second,
hundreds to thousands of times faster than existing detection systems. Our
system uses global image context to detect and localize objects, making it less
prone to background errors than top detection systems like R-CNN. By itself,
YOLO detects objects at unprecedented speeds with moderate accuracy. When
combined with state-of-the-art detectors, YOLO boosts performance by 2-3%
points mAP.
Can we efficiently learn the parameters of directed probabilistic models, in
the presence of continuous latent variables with intractable posterior
distributions, and in case of large datasets? We introduce a novel learning and
approximate inference method that works efficiently, under some mild
conditions, even in the on-line and intractable case. The method involves
optimization of a stochastic objective function that can be straightforwardly
optimized w.r.t. all parameters, using standard gradient-based optimization
methods.
The method does not require the typically expensive sampling loops per
datapoint required for Monte Carlo EM, and all parameter updates correspond to
optimization of the variational lower bound of the marginal likelihood, unlike
the wake-sleep algorithm. These theoretical advantages are reflected in
experimental results.
Let M(x) denote the expected value at level x of the response to a certain experiment. M(x) is assumed to be a monotone function of x but is unknown to the experimenter, and it is desired to find the solution of the equation , where is a given constant. We give a method for making successive experiments at levels in such a way that will tend to in probability.
Many engineers and scientists perform one-factor-at-a-time (OFAT) experiments. They will continue to do so until they understand the advantages of designed experiments over OFAT experiments, and until they learn to recognize OFAT experiments so they can avoid them. A very effective way to illustrate the advantages of designed experiments, and to show ways in which OFAT experiments present them-selves in real life, is to introduce real examples of OFAT experiments and then demonstrate why a designed experi-ment would have been better. Three engineering examples of OFAT experiments are presented, as well as designed ex-periments that would have been better. The three examples have been successfully used in an industrial workshop and can also be used in academic courses.
Roboflow (Version 1.0) [Software]. Computer Vision
- B Dwyer
Dwyer, B., N. J., H. T., et. al. Roboflow (Version 1.0) [Software]. Computer Vision.
YOLOv4: Optimal Speed and Accuracy of Object Detection
- A Bochkovskiy
- C.-Y. Wang
- H.-Y M Liao
Bochkovskiy, A., C.-Y. Wang, and H.-Y. M. Liao. YOLOv4: Optimal Speed and Accuracy of
Object Detection. 2020.
ADAM: A Method for Stochastic Optimization
- D P Kingma
- J Lei Ba
Kingma, D. P., and J. Lei Ba. ADAM: A Method for Stochastic Optimization. Presented at 3rd
International Conference for Learning Representations, San Diego, 2015.
Decoupled Weight Decay Regularization
- I Loshchilov
- F Hutter
Loshchilov, I., and F. Hutter. Decoupled Weight Decay Regularization. Presented at the 9 th
International Conference for Learning Representations, New Orleans, 2019.
On the Variance of the Adaptive Learning Rate and Beyond
- L Liu
- G Tech
- P He
- W Chen
- X Liu
- J Gao
- J Han
Liu, L., G. Tech, P. He, W. Chen, X. Liu, J. Gao, and J. Han. On the Variance of the Adaptive
Learning Rate and Beyond. Presented at the 10th International Conference for Learning
Representations, 2020.
Incorporating Nesterov Momentum into Adam
- Timothy Dozat
Timothy Dozat. Incorporating Nesterov Momentum into Adam. Presented at 4th International
Conference for Learning Representations, San Juan, 2016.