spacewalk01

Try training your model on crowd dataset

but by default, yolov7 uses half-precision setting for test while yolov5 uses fp32 precision. https://github.com/WongKinYiu/yolov7/blob/b8956dd5a5bcbb81c92944545ca03390c22a695f/test.py#L40

I got the same problem. Did you solve it?

Solved the problem after building opencv with opencv-contrib using cmake.

@jaybdub how to train a larger model with 416x416 or 608x608. The current model size is quiet small and cannot detect small objects.

I just finished implementing it in windows C++. https://github.com/batselem/Human_pose_detection

I will update its readme soon.

@thancaocuong added the detailed installation guide.

@brenoslinhares I have the same problem. I found that there is no code for generating images in output_images. 'output_images' is only mentioned in engine.py. I think the code is incomplete.

Thank you very much for a quick response! Yes but tail_corrupt_list contains corrupted triples. How did you label them? As I skim through it, it looks like all the triples...