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.
| Method | Backbone | Pretrain | Lr schd | box AP | mask AP | Download |
|---|---|---|---|---|---|---|
| Mask R-CNN | EfficientMod-s-Conv | ImageNet-1K | 1x | 42.1 | 38.5 | checkpoint & log |
| Mask R-CNN | EfficientMod-s | ImageNet-1K | 1x | 43.6 | 40.3 | checkpoint & log |
To evaluate EfficientMod + Mask R-CNN on COCO val2017, run:
dist_test.sh configs/{configure-file} /path/to/checkpoint_file 8 --out results.pkl --eval bbox segm
To train EfficientMod + Mask R-CNN on COCO train2017:
dist_train.sh configs/{configure-file} 8