When are Foundation Models Effective? Understanding the Suitability for Pixel-Level Classification Using Multispectral Imagery

TL;DR

This study evaluates the effectiveness of foundation models for pixel-level multispectral imagery classification, revealing they are not always superior to traditional methods and their suitability depends on task-specific factors.

Contribution

It provides a comparative analysis of foundation models versus traditional and deep learning models for remote sensing classification, highlighting when foundation models are appropriate.

Findings

Traditional ML models perform comparably or better in texture-insensitive tasks.

Deep learning models excel in texture-dependent tasks like burn scar detection.

Foundation models' effectiveness depends on the alignment of self-supervised tasks with real applications.

Abstract

Foundation models, i.e., very large deep learning models, have demonstrated impressive performances in various language and vision tasks that are otherwise difficult to reach using smaller-size models. The major success of GPT-type of language models is particularly exciting and raises expectations on the potential of foundation models in other domains including satellite remote sensing. In this context, great efforts have been made to build foundation models to test their capabilities in broader applications, and examples include Prithvi by NASA-IBM, Segment-Anything-Model, ViT, etc. This leads to an important question: Are foundation models always a suitable choice for different remote sensing tasks, and when or when not? This work aims to enhance the understanding of the status and suitability of foundation models for pixel-level classification using multispectral imagery at moderate resolution, through comparisons with traditional machine learning (ML) and regular-size deep learning models. Interestingly, the results reveal that in many scenarios traditional ML models still have similar or better performance compared to foundation models, especially for tasks where texture is less useful for classification. On the other hand, deep learning models did show more promising results for tasks where labels partially depend on texture (e.g., burn scar), while the difference in performance between foundation models and deep learning models is not obvious. The results conform with our analysis: The suitability of foundation models depend on the alignment between the self-supervised learning tasks and the real downstream tasks, and the typical masked autoencoder paradigm is not necessarily suitable for many remote sensing problems.