Are the saved pre-trained weights correct?

[lees12Penn]

I appreciate your efforts in sharing the codes and pre-trained model weights. However, after testing all three steps - heart localization, heart segmentation, and CAC segmentations, I've encountered issues. The pre-trained weights appear to be malfunctioning, and I suspect this could be due to modifications in the model architecture when the weights were saved, or possibly due to corruption of the weights file. For instance, when running heart localization with your provided model weight, I received the following error: "ValueError: You are trying to load a weight file containing 1 layer into a model with 19 layers." Could you please confirm that the pre-trained weights you've shared for all three steps are functioning as intended?

[emi-si]

Hello, I am facing the same issue. Did you manage to solve this problem?

[lees12Penn]

@emi-si: I've received no response and the problem remains unsolved.

[simmed00]

I encounter the same problem. Hope the author would help to clarify this

[Yinka890]

same issues, I cannot proceed without fixing the cause of the error. Has anyone found a way to diagnose the issue?

[Cardiate]

Dear colleagues,

Thank you for sharing the code and weights.

I’m trying to reproduce the results from the paper, but when I run "run_step1_heart_localization.py", I encounter the following error:

"ValueError: You are trying to load a weight file containing 1 layer into a model with 19 layers."

Could the AIM team kindly assist with this issue?