Deep View-Sensitive Pedestrian Attribute Inference in an end-to-end Model

TL;DR

This paper introduces an end-to-end view-sensitive model for pedestrian attribute inference, jointly predicting pedestrian pose and attributes, leading to improved accuracy in challenging surveillance datasets.

Contribution

It proposes a novel end-to-end framework that jointly predicts pedestrian view and attributes, enhancing attribute inference accuracy over existing methods.

Findings

Improved attribute prediction performance on PETA, RAP, and WIDER datasets.

View-sensitive model outperforms state-of-the-art methods.

Joint pose and attribute prediction benefits from shared learning.

Abstract

Pedestrian attribute inference is a demanding problem in visual surveillance that can facilitate person retrieval, search and indexing. To exploit semantic relations between attributes, recent research treats it as a multi-label image classification task. The visual cues hinting at attributes can be strongly localized and inference of person attributes such as hair, backpack, shorts, etc., are highly dependent on the acquired view of the pedestrian. In this paper we assert this dependence in an end-to-end learning framework and show that a view-sensitive attribute inference is able to learn better attribute predictions. Our proposed model jointly predicts the coarse pose (view) of the pedestrian and learns specialized view-specific multi-label attribute predictions. We show in an extensive evaluation on three challenging datasets (PETA, RAP and WIDER) that our proposed end-to-end view-aware attribute prediction model provides competitive performance and improves on the published state-of-the-art on these datasets.