AP-Loss for Accurate One-Stage Object Detection

Chen, Kean; Lin, Weiyao; Li, Jianguo; See, John; Ji, Wang

doi:10.1109/tpami.2020.2991457

Cited by 59 publications

(48 citation statements)

References 63 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DR Loss [31] achieves ranking between positives and negatives by enforcing a margin with Hinge Loss. Finally, AP Loss [6] and aLRP Loss [27] optimize the performance metrics, AP and LRP [26] respectively, by using the error-driven update of perceptron learning [35] for the non-differentiable parts. However, they need longer training and heavy augmentation.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Rank & Sort Loss for Object Detection and Instance Segmentation

Öksüz

Cam

Akbaş

et al. 2021

2021 IEEE/CVF International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

We propose Rank & Sort (RS) Loss, a ranking-based loss function to train deep object detection and instance segmentation methods (i.e. visual detectors). RS Loss supervises the classifier, a sub-network of these methods, to rank each positive above all negatives as well as to sort positives among themselves with respect to (wrt.) their localisation qualities (e.g. Intersection-over-Union -IoU). To tackle the non-differentiable nature of ranking and sorting, we reformulate the incorporation of error-driven update with backpropagation as Identity Update, which enables us to model our novel sorting error among positives. With RS Loss, we significantly simplify training: (i) Thanks to our sorting objective, the positives are prioritized by the classifier without an additional auxiliary head (e.g. for centerness, IoU, mask-IoU), (ii) due to its ranking-based nature, RS Loss is robust to class imbalance, and thus, no sampling heuristic is required, and (iii) we address the multi-task nature of visual detectors using tuning-free task-balancing coefficients. Using RS Loss, we train seven diverse visual detectors only by tuning the learning rate, and show that it consistently outperforms baselines: e.g. our RS Loss improves (i) Faster R-CNN by ∼ 3 box AP and aLRP Loss (rankingbased baseline) by ∼ 2 box AP on COCO dataset, (ii) Mask R-CNN with repeat factor sampling (RFS) by 3.5 mask AP ( ∼ 7 AP for rare classes) on LVIS dataset; and also outperforms all counterparts. Code is available at: https: //github.com/kemaloksuz/RankSortLoss.

show abstract

Section: Related Workmentioning

confidence: 99%

“…Recently proposed ranking-based loss functions, namely "Average Precision (AP) Loss" [6] and "average Localisation Recall Precision (aLRP) Loss" [27], offer two important advantages over the classical score-based functions (e.g. Cross-entropy Loss and Focal Loss [22]): (1) They directly optimize the performance measure (e.g.…”

Section: Introductionmentioning

confidence: 99%

Rank & Sort Loss for Object Detection and Instance Segmentation

Öksüz

Cam

Akbaş

et al. 2021

2021 IEEE/CVF International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

show abstract

“…The entire performance comparisons between the optimum InVision model and the 10 comparison methods are shown in Table 2. Particularly, InVision outperforms (1) the seven general DCNN-based approaches [33][34][35][36][37][38][39] by 23.1%-34.6% with respect to mAP and outperforms (2) the three DCNN-based ISAR recognition approaches described in [40][41][42] by 33.6%-37.1%. Moreover, all the compared methods take too long time to run (more than 18 s), while InVision costs only 0.607 s, which is 31.24 × 80.…”

Section: Overall Performance Evaluationmentioning

confidence: 99%

Robot target recognition using deep federated learning

Xue

Feng

et al. 2021

Int J of Intelligent Sys

View full text Add to dashboard Cite

Robot target recognition is a critical and fundamental machine vision task. In this paper, InVision, a robot target recognition approach is proposed using deep federated learning. Particularly, deep geometric learning is developed to improve the perception capabilities of convolutional neural networks, and promote the representation maps' resolutions while achieving good recognition performance. Moreover, federated metric learning is constructed to protect user data privacy across multiple devices and relieve the problem of inadequate available labeled training data. To improve the speed of the recognition system, a lightweight deep neural network is presented. Extensive experiments are performed, showing that InVision significantly outperforms the outstanding comparison approaches.

show abstract

“…AP-Loss [36] transforms the classification task into the sorting task and minimizes the AP-Loss of the system based on the network error and its optimization algorithm. Firstly, the prediction box and score are transformed to obtain the transformation format of the prediction box and score, as shown in the following equations:…”

Section: Optimization Of Loss Functionmentioning

confidence: 99%

An Apple Detection Method Based on Des-YOLO v4 Algorithm for Harvesting Robots in Complex Environment

Chen

Zhang

Bi-yu

et al. 2021

Mathematical Problems in Engineering

View full text Add to dashboard Cite

Real-time detection of apples in natural environment is a necessary condition for robots to pick apples automatically, and it is also a key technique for orchard yield prediction and fine management. To make the harvesting robots detect apples quickly and accurately in complex environment, a Des-YOLO v4 algorithm and a detection method of apples are proposed. Compared with the current mainstream detection algorithms, YOLO v4 has better detection performance. However, the complex network structure of YOLO v4 will reduce the picking efficiency of the robot. Therefore, a Des-YOLO structure is proposed, which reduces network parameters and improves the detection speed of the algorithm. In the training phase, the imbalance of positive and negative samples will cause false detection of apples. To solve the above problem, a class loss function based on AP-Loss (Average Precision Loss) is proposed to improve the accuracy of apple recognition. Traditional YOLO algorithm uses NMS (Nonmaximum Suppression) method to filter the prediction boxes, but NMS cannot detect the adjacent apples when they overlap each other. Therefore, Soft-NMS is used instead of NMS to solve the problem of missing detection, so as to improve the generalization of the algorithm. The proposed algorithm is tested on the self-made apple image data set. The results show that Des-YOLO v4 network has ideal features with a mAP (mean Average Precision) of apple detection of 97.13%, a recall rate of 90%, and a detection speed of 51 f/s. Compared with traditional network models such as YOLO v4 and Faster R-CNN, the Des-YOLO v4 can meet the accuracy and speed requirements of apple detection at the same time. Finally, the self-designed apple-harvesting robot is used to carry out the harvesting experiment. The experiment shows that the harvesting time is 8.7 seconds and the successful harvesting rate of the robot is 92.9%. Therefore, the proposed apple detection method has the advantages of higher recognition accuracy and faster recognition speed. It can provide new solutions for apple-harvesting robots and new ideas for smart agriculture.

show abstract

AP-Loss for Accurate One-Stage Object Detection

Cited by 59 publications

References 63 publications

Rank & Sort Loss for Object Detection and Instance Segmentation

Rank & Sort Loss for Object Detection and Instance Segmentation

Robot target recognition using deep federated learning

An Apple Detection Method Based on Des-YOLO v4 Algorithm for Harvesting Robots in Complex Environment

Contact Info

Product

Resources

About