Person Retrieval in Surveillance Video using Height, Color and Gender

TL;DR

This paper introduces a deep learning method for person retrieval in surveillance videos using soft biometric attributes like height, color, and gender, employing Mask R-CNN for segmentation and fine-tuned AlexNet models for attribute recognition.

Contribution

It presents a novel deep learning-based linear filtering approach that combines segmentation and attribute recognition for improved person retrieval in challenging surveillance scenarios.

Findings

Achieves high accuracy in person retrieval using semantic queries.

Effective segmentation with Mask R-CNN reduces background clutter.

Demonstrates robustness in challenging conditions on SoftBioSearch dataset.

Abstract

A person is commonly described by attributes like height, build, cloth color, cloth type, and gender. Such attributes are known as soft biometrics. They bridge the semantic gap between human description and person retrieval in surveillance video. The paper proposes a deep learning-based linear filtering approach for person retrieval using height, cloth color, and gender. The proposed approach uses Mask R-CNN for pixel-wise person segmentation. It removes background clutter and provides precise boundary around the person. Color and gender models are fine-tuned using AlexNet and the algorithm is tested on SoftBioSearch dataset. It achieves good accuracy for person retrieval using the semantic query in challenging conditions.