python main/train.py --config experiments/siamfcpp/train/got10k/siamfcpp_googlenet-trn.yaml -r /home/etvuz/projects/adversarial_attack/video_analyst/models/siamfcpp/siamfcpp-googlenet-got-md5_e182dc4c3823427022eccf7313d740a7.pkl
source activate pytorch1.4_python3.7
export CUDA_VISIBLE_DEVICES=1,2,3
cd /home/etvuz/projects/adversarial_attack/video_analyst
python main/train.py --config experiments/siamfcpp/train/fulldata/siamfcpp_googlenet-trn-fulldata.yaml -r /home/etvuz/projects/adversarial_attack/video_analyst/models/siamfcpp/siamfcpp-googlenet-got-md5_e182dc4c3823427022eccf7313d740a7.pkl
main/visualize_txt_result.py
videoanalyst.utils.visualize_training
videoanalyst.utils.visualize_search_img
生成相对于原始图像的 Fake Ground Truth
python videoanalyst/pipeline/utils/generate_patch_anno.py
数据位于 /home/etvuz/projects/adversarial_attack/patch_anno
[修改相应路径] python videoanalyst/pipeline/utils/generate_patch_anno.py
数据位于 /home/etvuz/projects/adversarial_attack/patch_anno/[dataset_name]
got_benchmark_helper.py/PipelineTracker/track():dataset_name = [dataset_name]
siamfcpp_track.py/self.do_attack = ...
siamfcpp_track.py/self.loop_num = ...
/home/etvuz/projects/adversarial_attack/video_analyst/models/siamfcpp/siamfcpp-googlenet-got-md5_e182dc4c3823427022eccf7313d740a7.pkl
source activate pytorch1.4_python3.7
cd /home/etvuz/projects/adversarial_attack/video_analyst
python main/test.py --config experiments/siamfcpp/train/got10k/siamfcpp_googlenet-trn.yaml
python main/test.py --config experiments/siamfcpp/test/otb/siamfcpp_googlenet-otb.yaml
python main/test.py --config experiments/siamfcpp/test/lasot/siamfcpp_googlenet-lasot.yaml
[修改 dataset_name/loop_num] python main/eval.py
python main/realtime_eval.py