WindVChen

哈喽 @SwjtuerFz ， concatFusionFactor 相比于 concat 只多了一个缩放系数，应该是不会影响程序的运行的，可以查查下其他代码段？

Maybe you can try this. (Just refer to the current `DRENet.yaml` template: remove the PANet part, replace C3 to C3ResAtnMHSA, add RCAN, modify the order index.) By the way, modify...

The error seems to come from the inconsistency between the output dimension of last module and the input dimension of next module. However, based on the current information, I have...

Have you modified the `parse_model( )` function in `yolo.py`? If you add an new SPPF class, you may add the class name the same as [here](https://github.com/ultralytics/yolov5/blob/789551ffd1ff4c43f53129454e39e5a35d1ec905/models/yolo.py#L318).

Hi @JunMa11 , Thanks for your provided slurm.sh. Based on it, I successfully adapted my codes into multi-node training. However, there is a strange phenomenon that doubts me a lot....

Hi there, sorry for replying late. There seems some useful tools for the yolo2voc conversion provided [here](https://stackoverflow.com/questions/68245781/how-to-convert-yolo-annotations-txt-to-pascal-voc-xml). You can use them for quick conversions.

Hi @ruffiann, Appreciate your interest in our work. Absolutely, we will release the code. As discussed in https://github.com/WindVChen/Diff-Harmonization/issues/1#issuecomment-1642412165, our aim is to have it out before September, barring any unexpected...

Hi @ruffiann , Exciting update! Our code is now live in the repository. Apologies for slightly missing the 'before September' target, but we made it just in the nick of...

哈喽 @yangjiuqian ， 如果指的是预测**无真值标签的图像**的话，一种直接的方法是，将所有待预测图像都放置于同一文件夹下，文件结构可以参考如下： ``` ├── test ├── 0 # 你可以更改为其他文件夹名 ├── img_1.jpg # 放置待预测图像 ├── img_2.jpg ├── ... ``` 之后参考[README.md](https://github.com/WindVChen/Push-and-Pull-Network#readme)中的Test Process即可作预测。 以上方法实际上是将所有图像归为假标签0，因此请忽略结果输出的Acc值。

哈喽呀 @liuz25 , 国庆快乐！ 从所给的报错信息来看，这可能是因为预训练权重没能正确导入，可以检查下模型加载路径是否正确.