Hi Dr Xie,
Thanks for your contribution on the topic of multispectral object detection. In your WACV 2024 paper, I found the results are surprisingly good. But I still have questions about the implementation details. For example, (1) what's the speed of your model, since it is built based on YOLOV7, which has diverse backbones? (2) Why the weights in loss are so effective, since the result is only 6.91 on KAIST when not using them in Table 3?
I think it would be much helpful if you can release the code and checkpoints for this paper. Thanks a lot!
Hi Dr Xie,
Thanks for your contribution on the topic of multispectral object detection. In your WACV 2024 paper, I found the results are surprisingly good. But I still have questions about the implementation details. For example, (1) what's the speed of your model, since it is built based on YOLOV7, which has diverse backbones? (2) Why the weights in loss are so effective, since the result is only 6.91 on KAIST when not using them in Table 3?
I think it would be much helpful if you can release the code and checkpoints for this paper. Thanks a lot!