train.sh

2019-02-13 本文已影响0人 ad丶leo

model_name="e2e_faster_rcnn_R-50-C4_1x"
mkdir -p ./logs/ $model_name DATE=`date '+%Y-%m-%d-%H-%M-%S'` CUDA_VISIBLE_DEVICES=0,1 python tools/train_net_step.py --dataset coco2014 \ --cfg configs/baselines/$ model_name.yaml
--load_detectron ./model.pkl
--use_tfboard --bs 4 --nw 4 --disp_interval 100
2>&1 | tee -a ./logs/ $model_name/$ {DATE}-logs.out