Autofocus Layer for Semantic Segmentation

Yao Qin, Konstantinos Kamnitsas, Siddarth Ancha, Jay Nanavati, Garrison Cottrell, Antonio Criminisi, Aditya Nori

Published 2018-05-22Version 1

We propose the autofocus convolutional layer for semantic segmentation with the objective of enhancing the capabilities of neural networks for multi-scale processing. Autofocus layers adaptively change the size of the effective receptive field based on the processed context to generate more powerful features. This is achieved by parallelising multiple convolutional layers with different dilation rates, combined by an attention mechanism that learns to focus on the optimal scales driven by context. By sharing the weights of the parallel convolutions we introduce only a small number of trainable parameters. The proposed autofocus layer can be easily integrated into existing networks to cope with biological variability among tasks. We evaluate our models on the challenging tasks of multi-organ segmentation in pelvic CT and brain tumor segmentation in MRI and achieve very promising performance.