Wang Ao

Thanks for your interest! The Adam optimizer is supported in this codebase. You could specify the `optimizer='Adam'` like YOLOv8. Besides, we would also like to confirm if you are using...

Thanks. We provide some clarification in #100 and https://huggingface.co/spaces/SkalskiP/YOLO-ARENA/discussions/1. > Moreover, we want to clarify that due to that YOLOv10 adopts the different training strategy with YOLOv8 and YOLOv9, it...

Thanks. Regarding the performance, could this comment https://github.com/THU-MIG/yolov10/issues/111#issuecomment-2137277092 help? Regarding the inference speed, could you please provide more details about the measurement? Thank you!

Thanks! Could you please try to use the exported format, e.g., TensorRT engine, to benchmark the speed? Due to that the one-to-many head, i.e., `cv2` and `cv3` of `v10Detect` is...

Please feel free to reopen this issue if you have further questions.

Thank you very much! We have updated README to highlight the onnx weights and the web demo.

@xenova Thanks for your great efforts! We found there are some differences between the predicted results in our local environment and in the huggingface demo. Here is a minimal example...

@xenova Thanks a lot for your great efforts! However, we find that the results between our local environment and the HuggingFace demo are still different. Specifically, we fork your fantastic...

@xenova We further investigated this and found that the reason may not be the input format in this HuggingFace demo. For onnx model, it's input format is `RGB`, as shown...

@xenova Thanks a lot for your great efforts! But we test the above image in the HuggingFace demo and the result is below, which is different from yours. Do we...