Keras-yolo3: Why is there no result with tiny-yolov3 prediction？Found 0 boxes for img为什么tiny-yolov3训练好之后，无法预测任何结果？一个boundingbox都出不来

[russellwest44]

After training the yolov3 network, the prediction is normal.but when training the tiny-yolov3,there was no bounding boxes.Everything follows the training steps. I successfully trained tiny-yolov3 for the first time and predicted the result, but this time it won’t work.

[wangbingqi]

hi,Have you solved this problem? I also encountered the same problem, tiny-yolov3 has no prediction result.

[russellwest44]

hi,Have you solved this problem? I also encountered the same problem, tiny-yolov3 has no prediction result.

Problem solved successfully！
There are some issues in the source code.

yolo3/model.py

anchor_mask = [[6,7,8], [3,4,5], [0,1,2]] if num_layers==3 else [[3,4,5], [1,2,3]]
please change all [1,2,3] to [0,1,2]
anchor_mask = [[6,7,8], [3,4,5], [0,1,2]] if num_layers==3 else [[3,4,5], [0,1,2]]
I remember there should be three mistakes
then retrain the tiny_model

[wangbingqi]

thank you

[dearwsj]

hi,Have you solved this problem? I also encountered the same problem, tiny-yolov3 has no prediction result.

Problem solved successfully！
There are some issues in the source code.

yolo3/model.py

anchor_mask = [[6,7,8], [3,4,5], [0,1,2]] if num_layers==3 else [[3,4,5], [1,2,3]]
please change all [1,2,3] to [0,1,2]
anchor_mask = [[6,7,8], [3,4,5], [0,1,2]] if num_layers==3 else [[3,4,5], [0,1,2]]
I remember there should be three mistakes
then retrain the tiny_model

请问是将cfg里面的mask也改成[[3,4,5], [0,1,2]]然后重新训练吗

[heshuju]

I have the same problem, When I find "[[3,4,5], [1,2,3]]", Let me test it. But why? I do not think it is a bug.

[tfukumori]

The original darknet was updated on August 16, 2018.

https://github.com/pjreddie/darknet/commit/f86901f6177dfc6116360a13cc06ab680e0c86b0#diff-2b0e16f442a744897f1606ff1a0f99d3 ...

However, this repository (qqwweee/keras-yolo3) was last updated on July 31, 2018, so the above changes do not appear to have been applied.