VisionISP: Repurposing the Image Signal Processor for Computer Vision Applications

TL;DR

VisionISP is a set of methods to adapt traditional image signal processors for computer vision tasks, reducing data size while maintaining relevant information, thereby enhancing object detection performance in autonomous driving.

Contribution

The paper introduces VisionISP, a novel approach to repurpose ISPs for machine vision, optimizing data for computer vision applications instead of human perception.

Findings

Significantly reduces data transmission requirements.

Improves object detection performance in autonomous driving.

Content-aware, trainable blocks enable effective adaptation.

Abstract

Traditional image signal processors (ISPs) are primarily designed and optimized to improve the image quality perceived by humans. However, optimal perceptual image quality does not always translate into optimal performance for computer vision applications. We propose a set of methods, which we collectively call VisionISP, to repurpose the ISP for machine consumption. VisionISP significantly reduces data transmission needs by reducing the bit-depth and resolution while preserving the relevant information. The blocks in VisionISP are simple, content-aware, and trainable. Experimental results show that VisionISP boosts the performance of a subsequent computer vision system trained to detect objects in an autonomous driving setting. The results demonstrate the potential and the practicality of VisionISP for computer vision applications.