A Comparison of Deep Learning Object Detection Models for Satellite Imagery

TL;DR

This paper compares various deep learning object detection models for satellite imagery, evaluating their accuracy and speed in detecting oil wells and small cars, highlighting trade-offs between model types.

Contribution

It provides a comprehensive comparison of single-stage, two-stage, and multi-stage models for satellite object detection, including performance benchmarks and insights for practical applications.

Findings

Single-stage detectors are faster and match accuracy for large objects.

Two-stage and multi-stage models are more accurate for small object detection.

Timing benchmarks establish a baseline for satellite object detection algorithms.

Abstract

In this work, we compare the detection accuracy and speed of several state-of-the-art models for the task of detecting oil and gas fracking wells and small cars in commercial electro-optical satellite imagery. Several models are studied from the single-stage, two-stage, and multi-stage object detection families of techniques. For the detection of fracking well pads (50m - 250m), we find single-stage detectors provide superior prediction speed while also matching detection performance of their two and multi-stage counterparts. However, for detecting small cars, two-stage and multi-stage models provide substantially higher accuracies at the cost of some speed. We also measure timing results of the sliding window object detection algorithm to provide a baseline for comparison. Some of these models have been incorporated into the Lockheed Martin Globally-Scalable Automated Target Recognition (GATR) framework.