Multi-Spectral Image Classification with Ultra-Lean Complex-Valued Models

TL;DR

This paper demonstrates that ultra-lean complex-valued deep learning models can effectively classify real-valued multi-spectral images, outperforming traditional methods even without data augmentation or transfer learning.

Contribution

It introduces the first application of complex-valued deep learning models to real-valued MSI data, showing superior performance with minimal model complexity.

Findings

Ultra-lean complex-valued models outperform ResNet on xView data.

Models trained from scratch without data augmentation achieve strong results.

First demonstration of complex-valued deep learning's effectiveness on real MSI images.

Abstract

Multi-spectral imagery is invaluable for remote sensing due to different spectral signatures exhibited by materials that often appear identical in greyscale and RGB imagery. Paired with modern deep learning methods, this modality has great potential utility in a variety of remote sensing applications, such as humanitarian assistance and disaster recovery efforts. State-of-the-art deep learning methods have greatly benefited from large-scale annotations like in ImageNet, but existing MSI image datasets lack annotations at a similar scale. As an alternative to transfer learning on such data with few annotations, we apply complex-valued co-domain symmetric models to classify real-valued MSI images. Our experiments on 8-band xView data show that our ultra-lean model trained on xView from scratch without data augmentations can outperform ResNet with data augmentation and modified transfer learning on xView. Our work is the first to demonstrate the value of complex-valued deep learning on real-valued MSI data.