pytorch_yolov1

pytorch_yolov1 copied to clipboard

我在运行train.sh的时候，调用的脚本是： python -m torch.distributed.launch --nproc_per_node 1 train_parallel.py 这个里面使用cuda运行的，而我没有英伟达的显卡，想用cpu跑，应该怎么修改python脚本呢？ 先谢谢了。

Kungdad

The build.py resize the image into (448,448), but i found the ground truth doesn't take the same operation. please hope you can answer it.

huangfuyb

Thanks for your work .I found the input box is like (x,y,sqrt(h),sqrt(y)),but the offset2box doesn't consider the sqrt .

huangfuyb

JAYCHOU2020

JAYCHOU2020

training confused, It's that normal????

1


GingerCohle

encoder

4

local_x = min(int(round((gt[2] + gt[0]) / 2)),w-1) local_y = min(int(round((gt[3] + gt[1]) / 2)),h-1) is this localtion error? gt = [2.1,3,3.3,1,4] ceil=(7,7) get local_x = 3 but the real local_x...

shitouren1994

您好，预训练好的模型的网址打不开

3

您好，预训练好的模型的网址打不开，能重新发一下预训练好的模型吗，感谢！

KaijieL