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.
✅ Main characteristics include:
-
Near real-time demonstration of person re-identification (ReID) task using OAK-D embedded vision platform and OpenVINO™ framework
-
Achieves ~10fps on Intel® Movidius™ Myriad™ X vision processor and OAK-D lite device (uses only color camera)**
-
Based on original project named: "Pedestrian reidentification" (MIT License Copyright (c) 2020 luxonis), original source code: https://github.com/luxonis/depthai-experiments/tree/master/gen2-pedestrian-reidentification
-
Modifications: new functionalities for visualization and control, parallel encoding of input camera feed and output video, recoding of output video with processing results of person re-identification
-
Requires the following pre-trained models (https://docs.openvino.ai/archives/index.html):
1) person-reidentification-retail-0288_openvino_2022.1_6shave
2) person-detection-retail-0013_openvino_2022.1_6shave
⭐* Hardware encoding of output video streams
⚡ * Test experiments: person re-identification in indoor and outdoor spaces under different conditions
📄 Publication preprint available at:
| 🚀 Feature | 📝 Specification |
|---|---|
| Processor | Intel Movidius Myriad X VPU |
| Cameras | 1x 13 MP (4208 x 3120) RGB (IMX214, rolling shutter) |
| 2x 0.31 MP (640 x 480) mono (OV7251, global shutter) | |
| Stereo baseline | 75 mm |
| AI performance | 4 TOPS (1.4 TOPS for AI) |
| Video output | Up to 4K@30fps (H.264/H.265/MJPEG) |
| Depth perception | Stereo, 300,000-point, 200+ FPS |
| Connectivity | USB 3.1 Gen1 Type-C |
| Dimensions | (WxHxD), 91 mm x 28 mm x 17.5 mm |
| Weight | 61 g |
| Variants | Fixed-focus / Auto-focus (RGB cam) |
| Mounting | 1/4”-20 tripod, VESA (7.5 cm, M4) |
(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.
(a) crowded indoor scene - successful ReID, but with low fps;
(b) false person detection, but with correct ReID;
(c) low light operation.
Presented implementation and experimental results are based on the pre-trained models kindly provided by the OpenVINO™ project: Open-source toolkit for optimizing and deploying AI inference.
The presented test videos were recorded for research purposes during live experiments and are provided under the Creative Commons Attribution-NonCommercial (CC BY-NC) license.
personReID is released under the MIT License terms in the provided LICENSE file.
[1] Brkljač, B., Brkljač, M. (2025). Person detection and re-identification in open-world settings of retail stores and public spaces. In Proceedings of the 2nd International Scientific Conference "ALFATECH – Smart Cities and modern technologies - 2025", Belgrade, Serbia, Feb. 28, 2025
@inproceedings{brkljacPersonReid2025,
author = {Brklja{\v{c}}, Branko and Brklja{\v{c}}, Milan},
title = {Person detection and re-identification in open-world settings of retail stores and public spaces},
booktitle = {Proceedings of the 2\textsuperscript{nd} International Scientific Conference ALFATECH – Smart Cities and modern technologies, Belgrade, Serbia},
volume = {1},
pages = {1--7},
month = {feb},
year = {2025},
doi = {-}
}
[2] Brkljač, B., Brkljač, M. (2025). Person detection and re-identification in open-world settings of retail stores and public spaces. arXiv preprint arXiv:2505.00772
@misc{brkljac2025persondetectionreidentificationopenworld,
title={Person detection and re-identification in open-world settings of retail stores and public spaces},
author={Branko Brkljač and Milan Brkljač},
year={2025},
eprint={2505.00772},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2505.00772},
doi={10.48550/arXiv.2505.00772}
}
