AAAI2020跟踪算法GlobalTrack的配置(GlobalTrack: A Simple and Strong Baseline for Long-term Tracking)

AAAI2020跟踪算法GlobalTrack的配置(GlobalTrack: A Simple and Strong Baseline for Long-term Tracking)_第1张图片

1、论文下载地址

GlobalTrack: A Simple and Strong Baseline for Long-term Tracking.[paper]

2、代码下载地址

https://github.com/huanglianghua/GlobalTrack

3、安装虚拟环境并激活

conda create -n GlobalTrack python=3.7
source activate GlobalTrack

4、安装依赖库

pip install torch==1.4.0 torchvision shapely scipy
pip install mmcv==0.2.14 matplotlib pycocotools terminaltables opencv-python-headless scikit-image

5、编译mmdetection

cd _submodules/mmdetection
python setup.py develop

6、预训练模型下载

Baidu Yun: [link] 百度网盘 请输入提取码 [password] 47p4

将下载的模型放到新建的pretrained_model路径中

AAAI2020跟踪算法GlobalTrack的配置(GlobalTrack: A Simple and Strong Baseline for Long-term Tracking)_第2张图片

 7、实验设置(仅以评估OTB2015为例)

打开tools/test_global_track.py

1)更改ckp_file为预训练模型所在路径,即第6步骤模型所存放的位置

2)更改evaluators,添加数据集所在路径

AAAI2020跟踪算法GlobalTrack的配置(GlobalTrack: A Simple and Strong Baseline for Long-term Tracking)_第3张图片

8、运行python tools/test_global_track.py

开始测试

AAAI2020跟踪算法GlobalTrack的配置(GlobalTrack: A Simple and Strong Baseline for Long-term Tracking)_第4张图片

9、可能遇到错误

cv2.error: OpenCV(4.5.3) /tmp/pip-req-build-fvfwe_ss/opencv/modules/highgui/src/window.cpp:1274: error: (-2:Unspecified error) The function is not implemented. Rebuild the library withWindows, GTK+ 2.x or Cocoa support. If you are on Ubuntu or Debian, install libgtk2.0-dev and pkg-config, then re-run cmake or configure script in function 'cvShowImage'

解决方法:

pip install opencv-contrib-python

你可能感兴趣的:(目标跟踪,目标跟踪,深度学习,计算机视觉,python)