@inproceedings{zhang2019freeanchor,
title = {{FreeAnchor}: Learning to Match Anchors for Visual Object Detection},
author = {Zhang, Xiaosong and Wan, Fang and Liu, Chang and Ji, Rongrong and Ye, Qixiang},
booktitle = {Neural Information Processing Systems},
year = {2019}
}
Backbone | Style | Lr schd | Mem (GB) | Inf time (fps) | box AP | Download |
---|---|---|---|---|---|---|
R-50 | pytorch | 1x | 4.9 | 18.4 | 38.7 | model | log |
R-101 | pytorch | 1x | 6.8 | 14.9 | 40.3 | model | log |
X-101-32x4d | pytorch | 1x | 8.1 | 11.1 | 41.9 | model | log |
Notes:
- We use 8 GPUs with 2 images/GPU.
- For more settings and models, please refer to the official repo.