Our detection implementation is based on MMDetection v2.19.0 and PVT detection. Thank the authors for their wonderful works.
Install MMDetection v2.19.0 from souce cocde,
or
pip install mmdet==2.19.0 --user
Apex (optional):
git clone https://github.com/NVIDIA/apex
cd apex
python setup.py install --cpp_ext --cuda_ext --user
Prepare COCO according to the guidelines in MMDetection v2.19.0.
| Backbone | Parmas | AP-box | AP-box@50 | AP-box@75 | AP-mask | AP-mask@50 | AP-mask@75 | Download |
|---|---|---|---|---|---|---|---|---|
| ResNet18 | 31.2M | 34.0 | 54.0 | 36.7 | 31.2 | 51.0 | 32.7 | |
| PoolFormer-S12 | 31.6M | 37.3 | 59.0 | 40.1 | 34.6 | 55.8 | 36.9 | |
| PVT-Tiny | 32.9M | 36.7 | 59.2 | 39.3 | 35.1 | 56.7 | 37.3 | |
| FCViT-B12 | 33.7M | 42.3 | 64.2 | 46.2 | 38.6 | 61.1 | 41.3 | [log & model] |
| ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- |
| ResNet50 | 44.2M | 38.0 | 58.6 | 41.4 | 34.4 | 55.1 | 36.7 | |
| Poolformer-S24 | 41.0M | 40.1 | 62.2 | 43.4 | 37.0 | 59.1 | 39.6 | |
| PVT-Small | 44.1M | 40.4 | 62.9 | 43.8 | 37.8 | 60.1 | 40.3 | |
| FCViT-B24 | 43.1M | 44.1 | 65.4 | 48.4 | 39.9 | 62.4 | 42.7 | [log & model] |
To evaluate FCViT-B12 + Mask R-CNN on COCO val2017, run:
dist_test.sh configs/mask_rcnn_fcvit_b12_fpn_1x_coco.py /path/to/checkpoint_file 8 --out results.pkl --eval bbox segm
To train FCViT-B12 + Mask R-CNN on COCO train2017:
dist_train.sh configs/mask_rcnn_fcvit_b12_fpn_1x_coco.py 8