Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Person detection and re-identification in open-world settings of retail stores and public spaces

Branko Brkljač, Milan Brkljač

TL;DR

The paper investigates open-world person detection and re-identification in retail stores and public spaces, emphasizing detection/localization as a prerequisite to ReID in multi-camera settings. It proposes a near real-time, cost-efficient solution on the OAK-D lite edge platform using pre-trained detection and ReID models via OpenVINO, and discusses hardware/software architecture and experimental validation. Experiments across indoor/outdoor environments reveal achievable frame rates around 12 fps with some reductions in challenging scenes, while orientation changes and background dynamics expose limitations in identity consistency and robustness. The work highlights future directions, including multimodal sensing and continuous model updates, to enhance robustness and applicability to marketing analytics while addressing privacy considerations.

Abstract

Practical applications of computer vision in smart cities usually assume system integration and operation in challenging open-world environments. In the case of person re-identification task the main goal is to retrieve information whether the specific person has appeared in another place at a different time instance of the same video, or over multiple camera feeds. This typically assumes collecting raw data from video surveillance cameras in different places and under varying illumination conditions. In the considered open-world setting it also requires detection and localization of the person inside the analyzed video frame before the main re-identification step. With multi-person and multi-camera setups the system complexity becomes higher, requiring sophisticated tracking solutions and re-identification models. In this work we will discuss existing challenges in system design architectures, consider possible solutions based on different computer vision techniques, and describe applications of such systems in retail stores and public spaces for improved marketing analytics. In order to analyse sensitivity of person re-identification task under different open-world environments, a performance of one close to real-time solution will be demonstrated over several video captures and live camera feeds. Finally, based on conducted experiments we will indicate further research directions and possible system improvements.

Person detection and re-identification in open-world settings of retail stores and public spaces

TL;DR

The paper investigates open-world person detection and re-identification in retail stores and public spaces, emphasizing detection/localization as a prerequisite to ReID in multi-camera settings. It proposes a near real-time, cost-efficient solution on the OAK-D lite edge platform using pre-trained detection and ReID models via OpenVINO, and discusses hardware/software architecture and experimental validation. Experiments across indoor/outdoor environments reveal achievable frame rates around 12 fps with some reductions in challenging scenes, while orientation changes and background dynamics expose limitations in identity consistency and robustness. The work highlights future directions, including multimodal sensing and continuous model updates, to enhance robustness and applicability to marketing analytics while addressing privacy considerations.

Abstract

Practical applications of computer vision in smart cities usually assume system integration and operation in challenging open-world environments. In the case of person re-identification task the main goal is to retrieve information whether the specific person has appeared in another place at a different time instance of the same video, or over multiple camera feeds. This typically assumes collecting raw data from video surveillance cameras in different places and under varying illumination conditions. In the considered open-world setting it also requires detection and localization of the person inside the analyzed video frame before the main re-identification step. With multi-person and multi-camera setups the system complexity becomes higher, requiring sophisticated tracking solutions and re-identification models. In this work we will discuss existing challenges in system design architectures, consider possible solutions based on different computer vision techniques, and describe applications of such systems in retail stores and public spaces for improved marketing analytics. In order to analyse sensitivity of person re-identification task under different open-world environments, a performance of one close to real-time solution will be demonstrated over several video captures and live camera feeds. Finally, based on conducted experiments we will indicate further research directions and possible system improvements.
Paper Structure (7 sections, 3 figures, 1 table)

This paper contains 7 sections, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related work
  3. Detection task
  4. Re-identification task
  5. Proposed person ReID implementation
  6. Experimental results and discussion
  7. Conclusions

Figures (3)

  • Figure 1: OAK-D lite OakDlite2022 vision platform: (a) three cameras, ISP and AI hardware accelerator on a single board, (b) implemented processing graph.
  • Figure 2: (a) succesful ReID, but with low fps rate due to dynamic background noise; (b) succesful ReID under low light conditions, but with identity loss after change of person's orientation at the end of sequence; (c) retail store application.
  • Figure 3: Experimental results visualization: (a) crowded indoor scene: successful ReID, but with low fps; (b) false person detection, but with correct ReID; (c) low light operation.