Hierarchical Attention Diffusion Networks with Object Priors for Video Change Detection

TL;DR

This paper introduces a hierarchical attention diffusion network that combines object priors and multi-scale attention to improve multi-class change detection in videos, achieving state-of-the-art results in remote sensing benchmarks.

Contribution

The paper proposes a unified change detection pipeline integrating instance masking, hierarchical cross attention, and semantic classification within a diffusion model, advancing the accuracy and interpretability of change maps.

Findings

Outperforms traditional methods by 10-25 points in F1 and IoU.

Effectively isolates novel objects using Mask R-CNN.

Achieves state-of-the-art results on synthetic and real-world datasets.

Abstract

We present a unified change detection pipeline that combines instance level masking, multi\-scale attention within a denoising diffusion model, and per pixel semantic classification, all refined via SSIM to match human perception. By first isolating only temporally novel objects with Mask R\-CNN, then guiding diffusion updates through hierarchical cross attention to object and global contexts, and finally categorizing each pixel into one of C change types, our method delivers detailed, interpretable multi\-class maps. It outperforms traditional differencing, Siamese CNNs, and GAN\-based detectors by 10\-25 points in F1 and IoU on both synthetic and real world benchmarks, marking a new state of the art in remote sensing change detection.