Intriguing Property and Counterfactual Explanation of GAN for Remote Sensing Image Generation

TL;DR

This paper investigates why GANs are more sensitive to training data size in remote sensing image generation, analyzes the phenomenon through experiments, and proposes regularization methods to improve feature information learning, leading to better generation quality.

Contribution

The paper introduces a causal model to interpret GAN data generation as counterfactuals and proposes two regularization schemes to enhance feature information learning in GANs for remote sensing images.

Findings

Regularization schemes improve GAN performance on RS datasets.

Theoretical proof links feature information to image quality.

Methods outperform existing models on multiple datasets.

Abstract

Generative adversarial networks (GANs) have achieved remarkable progress in the natural image field. However, when applying GANs in the remote sensing (RS) image generation task, an extraordinary phenomenon is observed: the GAN model is more sensitive to the size of training data for RS image generation than for natural image generation. In other words, the generation quality of RS images will change significantly with the number of training categories or samples per category. In this paper, we first analyze this phenomenon from two kinds of toy experiments and conclude that the amount of feature information contained in the GAN model decreases with reduced training data. Then we establish a structural causal model (SCM) of the data generation process and interpret the generated data as the counterfactuals. Based on this SCM, we theoretically prove that the quality of generated images is positively correlated with the amount of feature information. This provides insights for enriching the feature information learned by the GAN model during training. Consequently, we propose two innovative adjustment schemes, namely Uniformity Regularization (UR) and Entropy Regularization (ER), to increase the information learned by the GAN model at the distributional and sample levels, respectively. We theoretically and empirically demonstrate the effectiveness and versatility of our methods. Extensive experiments on three RS datasets and two natural datasets show that our methods outperform the well-established models on RS image generation tasks. The source code is available at https://github.com/rootSue/Causal-RSGAN.