基于siammask_e的成绩,验证下demo效果
https://paperswithcode.com/sota/visual-object-tracking-on-vot2019?p=fast-visual-object-tracking-with-rotated
安装步骤
1.下载项目
https://github.com/baoxinchen/siammask_e
git clone https://github.com/baoxinchen/siammask_e
2.Merge file to pysot 合并到pysot项目中,前提是已经配置好了PySOT项目
source activate pysot
cd siammask_e #切换到下载好的项目路径
bash install.sh [/path/to/pysot] # bash install.sh /home/cody/PycharmProjects/pysot
(pysot) cody@cody:~/PycharmProjects/siammask_e$ bash install.sh /home/cody/PycharmProjects/pysot
sending incremental file list
./
siammaske_r50_l3/
siammaske_r50_l3/config.yaml
sent 1,021 bytes received 46 bytes 2,134.00 bytes/sec
total size is 848 speedup is 0.79
sending incremental file list
./
core/
core/config.py
tracker/
tracker/siammask_e_tracker.py
tracker/tracker_builder.py
sent 18,175 bytes received 88 bytes 36,526.00 bytes/sec
total size is 17,835 speedup is 0.98
运行完毕之后,可以看到原有的pysot项目中有新添加的siammask_e文件夹,将siammask的model文件拷贝到siammask_e文件夹
3.测试siammask_E,过程与测试Pysot中其他模型文件一样
python tools/demo.py \
--config experiments/siammaske_r50_l3/config.yaml \
--snapshot experiments/siammaske_r50_l3/model.pth \
--video demo/bag.avi # (in case you don't have webcam)
4.效果