TOOD: Task-aligned One-stage Object Detection.

Chengjian Feng, Yujie Zhong, Yu Gao, Matthew R. Scott, Weilin Huang

Browse the full ICCV paper archive.