FluxML / FastAI.jl

Repository of best practices for deep learning in Julia, inspired by fastai
https://fluxml.ai/FastAI.jl
MIT License
585 stars 51 forks source link

Add `ConvFeatures` block to represent backbone outputs #278

Open lorenzoh opened 1 year ago

lorenzoh commented 1 year ago

Block representing features from a convolutional neural network backbone with n feature channels and N spatial dimensions.

For example, a 2D ResNet's convolutional layers may produce a hxwxch output that is passed further to the classifier head.

PR Checklist