You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

37 lines
1.4 KiB
Markdown

This file contains ambiguous Unicode characters!

This file contains ambiguous Unicode characters that may be confused with others in your current locale. If your use case is intentional and legitimate, you can safely ignore this warning. Use the Escape button to highlight these characters.

# YCB objects demo weights for Yolov4 & Yolov5
Any issuses please leave your comments at [YCB-Video dataset Download Mirror (YOLO labels & weights) | Nozomi](https://nozomi.sk/ycb-video-dataset-download-mirror/)
## Yolov4
1. Compile darknet.
2. Copy `yolov4/ycb` and `test1.mp4`to the folder of darknet.
![image-20210227202736516](https://i.loli.net/2021/02/28/eWtJM3p9sjAf27C.png)
3. Go to the folder of darknet, run `./darknet detector demo ycb/obj.data ycb/yolo-test.cfg ycb/weights/yolo-obj_7000.weights -thresh 0.5 -ext_output test1.mp4` .
4. For a headless server, you will meet an error (something looks like Gtk-WARNING cannot open display) if you execute line 3. Just try `./darknet detector demo ycb/obj.data ycb/yolo-test.cfg ycb/weights/yolo-obj_7000.weights -thresh 0.5 test1.mp4 -dont_show -ext_output -out_filename res.avi`
## Yolov5
1. Install yolov5 (assume you yolov5 project is located at `<yolov5_root>`)
2. Execute `mkdir -p <yolov5_root>/runs/train `
3. Copy `yolov5/exp2` to `<yolov5_root>/runs/train`
![image-20210227202532270](https://i.loli.net/2021/02/28/ioXfnA5jVdR2F8h.png)
4. Copy `test1.mp4` to `<yolov5_root>`
4. Run `python detect.py --source test1.mp4 --weights runs/train/exp2/weights/best.pt --conf-thres 0.25 --view-img`
## Test Videos
The videos `test1.mp4` and `test2.mp4` are collected from internet, **I dont own the rights to them**.