You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I try to follow directories to generate CAM. In your paper, you mentioned you used both GradCAM and GradCAM++, then decided to use GradCAM++. In your inference/cam_generation folder, there is generate_pascal.py file which I believe this file I should run to extract CAM. However, I see that you don't call either GradCAM and GradCAM++ in this file. Did you do some changes on GradCAM and used your own? Or should I go to another file?
Also, I am running the classification model (model_training/cam_generation/train_pascal.py) on a custom dataset by adjusting dataloaders. During training, I guess it is overfitting. My training loss is low whereas validation loss increases up to 33. My dataset is ~15.000 images. Did you also face that issue? Do you have any suggestions?
Lastly, am I referring correct directories? Thank you very much!
The text was updated successfully, but these errors were encountered:
Hello again,
I try to follow directories to generate CAM. In your paper, you mentioned you used both GradCAM and GradCAM++, then decided to use GradCAM++. In your inference/cam_generation folder, there is generate_pascal.py file which I believe this file I should run to extract CAM. However, I see that you don't call either GradCAM and GradCAM++ in this file. Did you do some changes on GradCAM and used your own? Or should I go to another file?
Also, I am running the classification model (model_training/cam_generation/train_pascal.py) on a custom dataset by adjusting dataloaders. During training, I guess it is overfitting. My training loss is low whereas validation loss increases up to 33. My dataset is ~15.000 images. Did you also face that issue? Do you have any suggestions?
Lastly, am I referring correct directories? Thank you very much!
The text was updated successfully, but these errors were encountered: