Enhancing Geometric Factors in Model Learning and Inference for Object Detection and Instance Segmentation

By Zhaohui Zheng, Ping Wang, Dongwei Ren, Wei Liu, Rongguang Ye, Qinghua Hu, Wangmeng Zuo. IEEE Transactions on Cybernetics, 2020

1,369 citations recorded by the source corpus.

Deep learning-based object detection and instance segmentation have achieved unprecedented progress. In this article, we propose complete-IoU (CIoU) loss and Cluster-NMS for enhancing geometric factors in both bounding-box regression and nonmaximum suppression (NMS), leading to notable gains of average precision (AP) and average recall (AR), without the sacrifice of inference efficiency. In particular, we consider three geometric factors, that is: 1) overlap area; 2) normalized central-point distance; and 3) aspect ratio, which are crucial for measuring bounding-box regression in object detection and instance segmentation. The three geometric factors are then incorporated into CIoU loss for better distinguishing difficult regression cases. The training of deep models using CIoU loss results in consistent AP and AR improvements in comparison to widely adopted $\ell _{n}$ -norm loss and IoU-based loss. Furthermore, we propose Cluster-NMS, where NMS during inference is done by implicitly clustering detected boxes and usually requires fewer iterations. Cluster-NMS is very efficient due to its pure GPU implementation, and geometric factors can be incorporated to improve both AP and AR. In the experiments, CIoU loss and Cluster-NMS have been applied to state-of-the-art instance segmentation (e.g., YOLACT and BlendMask-RT), and object detection (e.g., YOLO v3, SSD, and Faster R-CNN) models. Taking YOLACT on MS COCO as an example, our method achieves performance gains as +1.7 AP and +6.2 AR100 for object detection, and +1.1 AP and +3.5 AR100 for instance segmentation, with 27.1 FPS on one NVIDIA GTX 1080Ti GPU. All the source code and trained models are available at https://github.com/Zzh-tju/CIoU.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/63a06bd5f65a8d822a84905c370260e223cc78b2

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Yong Xu
2. Siliang Ma
3. Rongguang Ye
4. Wangmeng Zuo
5. Hongmin Shao
6. Jia Yao
7. Zhaohui Zheng
8. Zheng Liu

In Computer Vision and Pattern Recognition

1. Yong Xu
2. Siliang Ma
3. Dongwei Ren
4. Rongguang Ye
5. Wangmeng Zuo
6. Hongmin Shao
7. Jia Yao
8. Zhaohui Zheng

Citations in the corpus

• MPDIoU: A Loss for Efficient and Accurate Bounding Box Regression (2023, methodology): “Therefore, we try to design a novel loss function called L MPDIoU for bounding box regression with consideration of the advantages included in L GIoU [39], L DIoU [40], L CIoU [42], L EIoU [41], but also has higher e ffi ciency and accuracy for bounding box regression.”
• A Real-Time Detection Algorithm for Kiwifruit Defects Based on YOLOv5 (2021, background): “put forward the idea of DIoU and CIoU [33].”
• Object Detection in 20 Years: A Survey (2019, neutral): “object relationships and their spatial layout [118], [119].”
• Slim-neck by GSConv: a lightweight-design for real-time detector architectures (2022, background): “As research continues to evolve, more advanced IoU loss functions have been proposed, such as GIoU [34], DIoU [35], CIoU [36] and the EIoU [37].”
• Localization Distillation for Object Detection (2022, background): “In [57], [77], [85], [86], IoU-based loss functions are proposed to improve the localization quality of bounding boxes.”