realtime-semantic-segmentation-pytorch

realtime-semantic-segmentation-pytorch copied to clipboard

Have you used the pre-trained weights from the STDC official repository? If used, will mIoU be improved? Looking forward to your reply！

chris2000520

MIoU is less

6

Why is the miou of each of my models 10% less than the table? I ran 200 rounds of training.

chris2000520

1. Include custom dataset. 2. Implement .gitignore file. 3. Integrate requirements.txt. 4. Ensure compatibility with Segformer backbone. 5. Provide export onnx support for DDRNet and STDC.

acai66

Thank you for the nice code and comparsion of the real-time segmentation methods, Is the STDCseg is the best method? In other words, if i have a new segmentation job...

ezio1320

seg_trainer for using detail_loss may have bug

2

 detail seghead seems don't learn after a while and stay at same level

zhy168

  

zhy168

ljb-1

发现一处数值异常的地方，还请检查一下。

1

https://github.com/zh320/realtime-semantic-segmentation-pytorch?tab=readme-ov-file#full-resolution-on-cityscapes 这张表中，看到了DDRNET的FPS数有明显异常 233FPS，别的模型参数量比它小很多的也没有它快。 我找了官方的测试结果图  我们以BiSeNetv2作为参考，图中大概在150FPS，DDRnet的FPS在110左右。 而在你的表格中 BiSeNetv2 的FPS比较接近，而ddrnet的fps为233。我心生疑虑，还请勘误。

ZJDATY