search

taohan10200 / DRNet

PyTorch implementations of the paper: "DR.VIC: Decomposition and Reasoning for Video Individual Counting, CVPR, 2022"
50 stars 9 forks source link

How could I run the repo on a video and see the result? #7

Open munaiyi719 opened 1 year ago

munaiyi719 commented 1 year ago

How could I run the repo on a video and see the result?
I want to test a video like https://www.bilibili.com/video/BV1cY411H7hr/.

taohan10200 commented 1 year ago

We provide the demo_code in this repo, which includes how to fragment a video to frames and then inference on those frames. After that, you can reorgnized the predicted results to a video like the demo you referenced.

Hope this reply can help you!

munaiyi719 commented 1 year ago

I run video2img.py and run test_beijng.py. However I find that there is no Video_Crowd_Counting in model.VIC.

munaiyi719 commented 1 year ago

After I transform video to imgs, what need I do next?

taohan10200 commented 1 year ago

Please change the counter name to Video_Individual_Counter. After obtaining the imgs, you should modify the dir name in test_CroHD.py and then perform inference.