TY - GEN
T1 - Anchor box optimization for object detection
AU - Zhong, Yuanyi
AU - Wang, Jianfeng
AU - Peng, Jian
AU - Zhang, Lei
N1 - Publisher Copyright:
© 2020 IEEE.
PY - 2020/3
Y1 - 2020/3
N2 - In this paper, we propose a general approach to optimize anchor boxes for object detection. Nowadays, anchor boxes are widely adopted in state-of-the-art detection frameworks. However, these frameworks usually pre-define anchor box shapes in heuristic ways and fix the sizes during training. To improve the accuracy and reduce the effort of designing anchor boxes, we propose to dynamically learn the anchor shapes, which allows the anchors to automatically adapt to the data distribution and the network learning capability. The learning approach can be easily implemented with stochastic gradient descent and can be plugged into any anchor box-based detection framework. The extra training cost is almost negligible and it has no impact on the inference time or memory cost. Exhaustive experiments demonstrate that the proposed anchor optimization method consistently achieves significant improvement (≥ 1% mAP absolute gain) over the baseline methods on several benchmark datasets including Pascal VOC 07+12, MS COCO and Brainwash. Meanwhile, the robustness is also verified towards different anchor initialization methods and the number of anchor shapes, which greatly simplifies the problem of anchor box design.
AB - In this paper, we propose a general approach to optimize anchor boxes for object detection. Nowadays, anchor boxes are widely adopted in state-of-the-art detection frameworks. However, these frameworks usually pre-define anchor box shapes in heuristic ways and fix the sizes during training. To improve the accuracy and reduce the effort of designing anchor boxes, we propose to dynamically learn the anchor shapes, which allows the anchors to automatically adapt to the data distribution and the network learning capability. The learning approach can be easily implemented with stochastic gradient descent and can be plugged into any anchor box-based detection framework. The extra training cost is almost negligible and it has no impact on the inference time or memory cost. Exhaustive experiments demonstrate that the proposed anchor optimization method consistently achieves significant improvement (≥ 1% mAP absolute gain) over the baseline methods on several benchmark datasets including Pascal VOC 07+12, MS COCO and Brainwash. Meanwhile, the robustness is also verified towards different anchor initialization methods and the number of anchor shapes, which greatly simplifies the problem of anchor box design.
UR - http://www.scopus.com/inward/record.url?scp=85085468271&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85085468271&partnerID=8YFLogxK
U2 - 10.1109/WACV45572.2020.9093498
DO - 10.1109/WACV45572.2020.9093498
M3 - Conference contribution
AN - SCOPUS:85085468271
T3 - Proceedings - 2020 IEEE Winter Conference on Applications of Computer Vision, WACV 2020
SP - 1275
EP - 1283
BT - Proceedings - 2020 IEEE Winter Conference on Applications of Computer Vision, WACV 2020
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2020 IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2020
Y2 - 1 March 2020 through 5 March 2020
ER -