Closed ZiyeHu closed 3 years ago
You can follow this issue. https://github.com/ucbdrive/few-shot-object-detection/issues/87#issuecomment-768988092 You can also use your own pretrained weight by changing MODEL.WEIGHTS in yaml.
pip install fvcore==0.1.2.post20201020
will address the problem.
python tools/train_net.py --num-gpus 8 \ --config-file configs/PASCAL_VOC/base-training/R101_FPN_base_training_split1.yml