CDN-MEDAL: Two-stage Density and Difference Approximation Framework for Motion Analysis

TL;DR

CDN-MEDAL combines statistical Gaussian Mixture Models with deep neural networks in a two-stage framework to improve motion analysis in video surveillance, achieving rapid convergence and effective motion region extraction.

Contribution

It introduces a novel two-stage framework integrating GMM-based statistical learning with lightweight deep neural networks for efficient background subtraction.

Findings

Effective in extracting moving object regions in unseen scenarios

Rapid convergence to complex motion patterns

High efficiency in background modeling and subtraction

Abstract

Background modeling and subtraction is a promising research area with a variety of applications for video surveillance. Recent years have witnessed a proliferation of effective learning-based deep neural networks in this area. However, the techniques have only provided limited descriptions of scenes' properties while requiring heavy computations, as their single-valued mapping functions are learned to approximate the temporal conditional averages of observed target backgrounds and foregrounds. On the other hand, statistical learning in imagery domains has been a prevalent approach with high adaptation to dynamic context transformation, notably using Gaussian Mixture Models (GMM) with its generalization capabilities. By leveraging both, we propose a novel method called CDN-MEDAL-net for background modeling and subtraction with two convolutional neural networks. The first architecture, CDN-GM, is grounded on an unsupervised GMM statistical learning strategy to describe observed scenes' salient features. The second one, MEDAL-net, implements a light-weighted pipeline of online video background subtraction. Our two-stage architecture is small, but it is very effective with rapid convergence to representations of intricate motion patterns. Our experiments show that the proposed approach is not only capable of effectively extracting regions of moving objects in unseen cases, but it is also very efficient.