Get features from Mask R-CNN object

vision
1

Hi, as in this post I want to get the feature vector prior to the classification layer of the final detected objects (final output of Mask R-CNN). How could this be done with the pytorch implementation of Mask R-CNN?
Thanks!

2

Is the suggestion from the linked topic not working?
If so, could you explain your use case a bit and how it differs from theirs? :slight_smile: