Optimized clothes segmentation to boost gender classification in unconstrained scenarios

TL;DR

This paper presents a novel clothes segmentation method using trixels and an adapted GrabCut algorithm, significantly improving gender classification accuracy in unconstrained scenarios.

Contribution

It introduces trixels for image region clustering and modifies GrabCut for unsupervised clothes segmentation, enhancing gender classification performance.

Findings

Trixels outperform superpixels in segmentation tasks.

Modified GrabCut enables near real-time clothes segmentation.

Fusion of clothes features with gender classifiers improves accuracy.

Abstract

Several applications require demographic information of ordinary people in unconstrained scenarios. This is not a trivial task due to significant human appearance variations. In this work, we introduce trixels for clustering image regions, enumerating their advantages compared to superpixels. The classical GrabCut algorithm is later modified to segment trixels instead of pixels in an unsupervised context. Combining with face detection lead us to a clothes segmentation approach close to real time. The study uses the challenging Pascal VOC dataset for segmentation evaluation experiments. A final experiment analyzes the fusion of clothes features with state-of-the-art gender classifiers in ClothesDB, revealing a significant performance improvement in gender classification.