Task-Driven Super Resolution: Object Detection in Low-resolution Images

TL;DR

This paper introduces a task-driven super-resolution framework that jointly optimizes super-resolution and object detection, significantly enhancing detection accuracy in low-resolution images.

Contribution

It proposes an end-to-end training method that incorporates detection loss into super-resolution, improving object detection performance on low-res images.

Findings

Task-driven SR improves detection accuracy.

End-to-end training outperforms independent optimization.

Framework is effective across various conditions.

Abstract

We consider how image super resolution (SR) can contribute to an object detection task in low-resolution images. Intuitively, SR gives a positive impact on the object detection task. While several previous works demonstrated that this intuition is correct, SR and detector are optimized independently in these works. This paper proposes a novel framework to train a deep neural network where the SR sub-network explicitly incorporates a detection loss in its training objective, via a tradeoff with a traditional detection loss. This end-to-end training procedure allows us to train SR preprocessing for any differentiable detector. We demonstrate that our task-driven SR consistently and significantly improves accuracy of an object detector on low-resolution images for a variety of conditions and scaling factors.