nearly realtime:avg_fps=14 for mx150(2G GPU)
Weights files and demo.mp4 at here:(also i put this video to bilibili,after passing the audit,i will put 链接) [BaiduDisk] 链接：https://pan.baidu.com/s/1czw0dz56ZjfgrxjBFaUSdg 提取码：odbu Yolov3_tiny_1cls_se detection results at here: 链接：https://www.bilibili.com/video/av94365366/
- python 3
- numpy
- scipy
- opencv-python
- sklearn
- torch ==1.1.0
- torchvision ==0.3.0
- pillow
- vizer
- edict
-
Check all dependencies installed
-
Download best.pt(yolo weights) parameters
look up
- Download deepsort parameters ckpt.t7
look up and put here:
cd deep_sort/deep/checkpoint
Notice:
If compiling failed, the simplist way is to **Upgrade your pytorch >= 1.1 and torchvision >= 0.3" and you can avoid the troublesome compiling problems which are most likely caused by either gcc version too low or libraries missing.
- Run my_deep_sort.py
my is windows
i do not know linux or ubuntu
my programming ability is very weak
comments are not very good
forgive me
Notice: Maybe some mistakes, you can discuss in issues, help me to modify. It is just a small work, you can improve based on my code.
-
paper: Simple Online and Realtime Tracking with a Deep Association Metric
-
code: nwojke/deep_sort
-
paper: YOLOv3
-
code: ultralytics/yolov3