XiongZhongxia

Actually, I also implement a light-weight model (which nearly has 10% FLOPS of your baseline), with multispectral inputs. Finally it only gets 62.7 AP but achieves 4.3 MR. Is this...

> 1. A very interesting and valuable question. The following is my superficial understanding of this issue, there may be something wrong, and corrections are welcome. > According to section...

当x[:, 5]也就是坐标为0时，表示是未标注关键点，这一点你可以参考一下coco人体关键点标注的格式，这样的话labels[:, 5]就是0-1=-1，于是在计算loss时，label=-1的点loss会被屏蔽掉，可以在utils/loss.py第97行，weight[torch.where(t==-1)]=0看到这个设置。

http://box.vicos.si/vot/vot2013.zip There is an available link.

Putting line247-254 in the loop of each epoch will solve this.

> > The same question confuses me. I look forward to getting an answer. > > The author provides a pre-trained checkpoint named "yolov5l_transformerx3_llvip_s1024_bs32_e200". As [this](https://github.com/DocF/multispectral-object-detection/issues/4#issuecomment-979777484) reply mentioned, the author...

> Thanks for your attention. The dataset is available at https://zhang-pengyu.github.io/DUT-VTUAV/ Actually I asked for full dataset link because the provided links are for data splits... Could you please provide...

> Sorry, we cannot provide a full dataset link because of file size limitation. For convenience, we will provide a bash/python file for downloading the full dataset automatically. Thanks for...

> > Sorry, we cannot provide a full dataset link because of file size limitation. For convenience, we will provide a bash/python file for downloading the full dataset automatically. Thanks...