convalution havn't activate linear

[hufangjian]

你好用了你百度网盘上的yolov4.pth ,运行demo.py 报“convalution havn't activate linear“，并且无输出。

[Tianxiaomo]

“convalution havn't activate linear“不影响正常运行

[yilawu]

“convalution havn't activate linear“不影响正常运行
我和上面的兄弟问题一样，运行demo.py报convalution havn't activate linear
而且输出没有进行预测，就是bboxes是空的

[tsminh]

"Convalution havn't activate linear" does not affect the normal operation.
I have the same problem as the brothers above. Running demo.py reports the contribution havn't activate linear
and the output is not predicted, that is, the bboxes are empty.

Also me, i run the demo.py, and bbs are all empty

[yilawu]

"Convalution havn't activate linear" does not affect the normal operation.
I have the same problem as the brothers above. Running demo.py reports the contribution havn't activate linear
and the output is not predicted, that is, the bboxes are empty.

Also me, i run the demo.py, and bbs are all empty

I find the reason.
when you run the demo.py, you should use Darknet weights and
when you run the modules.py, you should use pytorch weights.
Don't be confused.

[3wh1te]

“convalution havn't activate linear“不影响正常运行，这是正常的输出吧，pytorch的提醒？

[dlrookies]

"Convalution havn't activate linear" does not affect the normal operation.
I have the same problem as the brothers above. Running demo.py reports the contribution havn't activate linear
and the output is not predicted, that is, the bboxes are empty.

Also me, i run the demo.py, and bbs are all empty

I find the reason.
when you run the demo.py, you should use Darknet weights and
when you run the modules.py, you should use pytorch weights.
Don't be confused.

can not find the modules.py

[xiecookie]

"Convalution havn't activate linear" does not affect the normal operation.
I have the same problem as the brothers above. Running demo.py reports the contribution havn't activate linear
and the output is not predicted, that is, the bboxes are empty.

Also me, i run the demo.py, and bbs are all empty

I find the reason.
when you run the demo.py, you should use Darknet weights and
when you run the modules.py, you should use pytorch weights.
Don't be confused.

请问现在你有输出了吗，也就是bbox是不是为空呢

[glee1228]

"Convalution havn't activate linear" does not affect the normal operation.
I have the same problem as the brothers above. Running demo.py reports the contribution havn't activate linear
and the output is not predicted, that is, the bboxes are empty.

Also me, i run the demo.py, and bbs are all empty

I find the reason.
when you run the demo.py, you should use Darknet weights and
when you run the modules.py, you should use pytorch weights.
Don't be confused.

can not find the modules.py

You can currently run models.py. Load the .pth file. This will not detect empty bboxes.

[cuihangmver]

我在python3.7，pytorch1.4，opencv4.5的环境中也遇到同样问题，并且我测试只有使用yolov4-tiny.cfg可以得到这个结果，其他两个文件yolov4.cfg，yolov4-custom.cfg得到错误：RuntimeError: shape '[256, 128, 3, 3]' is invalid for input of size 28044

[zdaiot]

see https://github.com/Tianxiaomo/pytorch-YOLOv4/blob/master/tool/darknet2pytorch.py#L274