环境

torch1.5+cu 101+cp38 on 2080ti

# clone the repository inclduing Detectron2(@be792b9) 
$ git clone --recursive https://github.com/MILVLG/bottom-up-attention.pytorch

$ cd detectron2
$ pip install -e .
$ cd ..

detectron2直接克隆有问题，需要把detecron2/detecron2中的所有文件放入detecron2下

git clone https://github.com/ptrblck/apex.git
cd apex
pip install -v --no-cache-dir ./


apt-get install libglib2.0-dev

 Faster R-CNN-k10-100 is the default model地址https://awma1-my.sharepoint.com/:u:/g/personal/yuz_l0_tn/EaXvCC3WjtlLvvEfLr3oa8UBLA21tcLh4L8YLbYXl6jgjg?download=1

python3 extract_features.py --mode caffe \
         --num-cpus 0 --gpus 1 \
         --extract-mode roi_feats \
         --min-max-boxes '36' \
         --config-file configs/caffe/test-caffe-r101.yaml \
         --image-dir /hy-tmp/img --bbox-dir /hy-tmp/out --out-dir /hy-tmp/out