Rock Hunting With Martian Machine Vision

TL;DR

This paper explores deep learning techniques for Martian rock detection using computer vision, achieving high accuracy in classification and developing low-power models suitable for onboard rover deployment.

Contribution

It introduces a method for classifying and detecting Martian rocks with deep learning, including model optimization for microcontroller use.

Findings

Over 97% accuracy in binary rock vs. rover classification

A detector with geo-located bounding boxes for rock counting

A low-power, quantized model operating at 1 fps with 37% accuracy

Abstract

The Mars Perseverance rover applies computer vision for navigation and hazard avoidance. The challenge to do onboard object recognition highlights the need for low-power, customized training, often including low-contrast backgrounds. We investigate deep learning methods for the classification and detection of Martian rocks. We report greater than 97% accuracy for binary classifications (rock vs. rover). We fine-tune a detector to render geo-located bounding boxes while counting rocks. For these models to run on microcontrollers, we shrink and quantize the neural networks' weights and demonstrate a low-power rock hunter with faster frame rates (1 frame per second) but lower accuracy (37%).