Smart City Gnosys

Smart city article details

Title	Multi-Task Model With Attribute-Specific Heads For Person Re-Identification
ID_Doc	38442
Authors	Ahmed M.F.; Oyshee A.A.N.
Year	2025
Published	Pattern Analysis and Applications, 28, 1
DOI	http://dx.doi.org/10.1007/s10044-025-01421-0
Abstract	Person re-identification (ReID) has become an important task in digital surveillance for enhancing security, efficient monitoring, and enabling various applications in smart cities and public safety systems. Person ReID with attributes is a challenging task due to different camera views create significant difficulties in capturing each person’s unique identity and detailed attributes. In this work, we propose a multi-task model that not only performs unique person ReID but also simultaneously predicts attributes. Our model jointly utilizes a shared backbone network, which can be either ResNet50 or EfficientNet, along with generalized mean (GeM) pooling to achieve efficient feature extraction. It also applies attribute-specific heads to predict various characteristics such as gender, age, type of clothes, color, and alongside the ReID classification. This multi-task approach utilizes the shared features across tasks, gives comprehensive attribute predictions, and may further contribute to identification in surveillance scenarios. We evaluate our model on two commonly used publicly available datasets, Market1501 and DukeMTMC-reID, demonstrating how our approach can improve both in ReID accuracy and give reliable attribute predictions. These results reveal that our multi-task model can be competitive, providing a holistic solution for practical applications in surveillance where both identification and attributes are important. The approach has shown the potential of unifying ReID with attribute prediction to develop more robust and advanced surveillance systems. The code of this experiment is publicly accessible at https://github.com/TripleTheGreatDali/ReIDMTMASH. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.
Author Keywords	Attribute head; Digital surveillance; Multi-task learning; Person re-identification

Similar Articles

Id	Similarity	Authors	Title	Published
18110	0.898	Khan S.U.; Hussain T.; Ullah A.; Baik S.W.	Deep-Reid: Deep Features And Autoencoder Assisted Image Patching Strategy For Person Re-Identification In Smart Cities Surveillance	Multimedia Tools and Applications, 83, 5 (2024)
41889	0.894	Qi L.; Ren Z.; Liu J.; Geng X.	Person Re-Identification Method Based On Hybrid Real-Synthetic Data; [虚实结合的行人重识别方法]	Jisuanji Yanjiu yu Fazhan/Computer Research and Development, 62, 2 (2025)
41892	0.882	Stojanovic D.; Pavkovic B.; Cetic N.	Person Re-Identification With Transformers And Image Stacking	2024 32nd Telecommunications Forum, TELFOR 2024 - Proceedings of Papers (2024)
51477	0.87	Perwaiz N.; Fraz M.M.; Shahzad M.	Smart Surveillance With Simultaneous Person Detection And Re-Identification	Multimedia Tools and Applications, 83, 5 (2024)
42375	0.867	HE Z.; ZHAO H.; WANG J.; FENG W.	Pose Matters: Pose Guided Graph Attention Network For Person Re-Identification	Chinese Journal of Aeronautics, 36, 5 (2023)
4278	0.856	Selvan C.; Anwar Basha H.; Meenakshi K.; Naveen S.	A Review On Person Re-Identification Techniques And Its Analysis	IEEE Access, 13 (2025)
27017	0.855	Zhuang J.; Wang N.; Zhuang Y.; Hao Y.	Frame Extraction Person Retrieval Framework Based On Improved Yolov8S And The Stage-Wise Clustering Person Re-Identification	IET Image Processing, 19, 1 (2025)
10212	0.855	Mamedov T.; Kuplyakov D.; Konushin A.	Approaches To Improve The Quality Of Person Re-Identification For Practical Use	Sensors, 23, 17 (2023)
39446	0.854	Cui L.; Liu M.; Wang R.; Zafar N.	Novel Attention-Based Framework For Person Re-Identification In Video Surveillance	Lecture Notes in Electrical Engineering, 1209 LNEE (2024)
6289	0.852	Sun H.; Zhao H.; Zhang W.; Xu L.; Guan H.	Adaptive Multi-Task Learning For Multi-Par In Real World	IEEE Journal of Radio Frequency Identification, 8 (2024)