Open igor-volgin opened 4 months ago
@igor-volgin Thanks for using this tool, the published models and .exe are initial versions, but the infer.py from mPSAT-backend is prepared for the newer ones.
I apologize for the absence of latest pretrained models, and I have published them at https://github.com/zhengGroupDEV/mPSAT-backend/releases/tag/v0.2.0 , please use the new .onnx models and try again.
Example:
Also, I recommand you to use CNN1D instead of CNN2D for it's better accuracy, if you need batch processing, you can modify the infer.py and implement it, hoping this will help you.
Dear developers,
I have tried to perform classification by running infer.py for cnn2d (the script was taken from mPSAT-backend/tree/master/src, and onnx file for cnn2d from Release). However, the results produced by the script were found to differ from those which I get in GUI for PU.csv. What is the reason?
Could you please give an advise on how to apply your method out-of-GUI to get similar results on PU.csv? I'm interested in testing your method on large number of files.
Thank you in advance!