Learning Multi-scale Features for Foreground Segmentation

TL;DR

This paper introduces a novel encoder-decoder neural network with multi-scale feature fusion for foreground segmentation, achieving state-of-the-art results with minimal training data.

Contribution

It extends the FgSegNet's Feature Pooling Module to incorporate internal feature fusions, enabling robust multi-scale feature extraction without multi-scale inputs.

Findings

Outperforms existing methods on CDnet2014 with F-Measure of 0.9847

Effective on SBI2015 and UCSD datasets

Requires only few training examples

Abstract

Foreground segmentation algorithms aim segmenting moving objects from the background in a robust way under various challenging scenarios. Encoder-decoder type deep neural networks that are used in this domain recently perform impressive segmentation results. In this work, we propose a novel robust encoder-decoder structure neural network that can be trained end-to-end using only a few training examples. The proposed method extends the Feature Pooling Module (FPM) of FgSegNet by introducing features fusions inside this module, which is capable of extracting multi-scale features within images; resulting in a robust feature pooling against camera motion, which can alleviate the need of multi-scale inputs to the network. Our method outperforms all existing state-of-the-art methods in CDnet2014 dataset by an average overall F-Measure of 0.9847. We also evaluate the effectiveness of our method on SBI2015 and UCSD Background Subtraction datasets. The source code of the proposed method is made available at https://github.com/lim-anggun/FgSegNet_v2 .