Skip to content
/ GCR Public

The implement of “ClickTrack: Towards Real-time Interactive Single Object Tracking”

Notifications You must be signed in to change notification settings

kuiran/GCR

Repository files navigation

Installation

Our code is based on MMtracking master branch Link, you can fllow the installation guiding installation to create your environment.

Training & Inference

Training

bash ./tools/dist_train.sh ${config_path} 8 --work-dir ${dir}
bash ./tools/dist_test.sh ${config_path} 8 \
 --checkpoint {model_path} \
 --out {work_dir}/result.pkl \
 --work-dir {work_dir} \

Inference

conver pkl to txt

 python ./utils/pkl2txt.py \
 --pkl_path {work_dir}/result.pkl \
 --txt_save_path {work_dir}/ \
 --txt_name result.txt

 bash ./tools/dist_test.sh {config_path} 8 \
 --checkpoint {model_path} \
 --out {work_dir}/point_results.pkl \
 --eval track\
 --work-dir ${work_dir}stark_st1_r50_500e_lasot \
 --cfg-options data.test.replace_first_frame_ann=True \
 data.test.first_frame_ann_path={work_dir}/result.txt

If you have any problem when using the code, please contact with us.

About

The implement of “ClickTrack: Towards Real-time Interactive Single Object Tracking”

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published