Spatially Adaptive Computation Time for Residual Networks

TL;DR

This paper introduces a Residual Network architecture that adaptively adjusts computation per image region, improving efficiency and aligning well with human visual attention across various vision tasks.

Contribution

It presents a novel spatially adaptive computation method for Residual Networks that is end-to-end trainable and applicable to multiple computer vision tasks.

Findings

Improved computational efficiency on ImageNet and COCO datasets.

Computation time maps correlate with human eye fixations.

Applicable to diverse vision problems without modifications.

Abstract

This paper proposes a deep learning architecture based on Residual Network that dynamically adjusts the number of executed layers for the regions of the image. This architecture is end-to-end trainable, deterministic and problem-agnostic. It is therefore applicable without any modifications to a wide range of computer vision problems such as image classification, object detection and image segmentation. We present experimental results showing that this model improves the computational efficiency of Residual Networks on the challenging ImageNet classification and COCO object detection datasets. Additionally, we evaluate the computation time maps on the visual saliency dataset cat2000 and find that they correlate surprisingly well with human eye fixation positions.