Generalising without Forgetting for Lifelong Person Re-Identification

By Guile Wu, Shaogang Gong. 2021

64 citations recorded by the source corpus.

Existing person re-identification (Re-ID) methods mostly prepare all training data in advance, while real-world Re-ID data are inherently captured over time or from different locations, which requires a model to be incrementally generalised from sequential learning of piecemeal new data without forgetting what is already learned. In this work, we call this lifelong person Re-ID, characterised by solving a problem of unseen class identification subject to continuous new domain generalisation and adaptation with class imbalanced learning. We formulate a new Generalising without Forgetting method (GwFReID) for lifelong Re-ID and design a comprehensive learning objective that accounts for classification coherence, distribution coherence and representation coherence in a unified framework. This design helps to simultaneously learn new information, distil old knowledge and solve class imbalance, which enables GwFReID to incrementally improve model generalisation without catastrophic forgetting of what is already learned. Extensive experiments on eight Re-ID benchmarks, CIFAR-100 and ImageNet show the superiority of GwFReID over the state-of-the-art methods.

Subfields: Biomedical Engineering, Computer Vision and Pattern Recognition, Mechanical Engineering

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

1. Yadong Mu
2. Zhicheng Sun
3. Pan Li
4. Guile Wu
5. Shaogang Gong
6. Liang Zheng
7. Yi Yang
8. Xiatian Zhu

In Computer Vision and Pattern Recognition

1. Yadong Mu
2. Zhicheng Sun
3. Pan Li
4. Guile Wu
5. Shaogang Gong
6. Jiahuan Zhou
7. Yuxin Peng
8. Zhenyu Cui

In Mechanical Engineering

1. Yadong Mu
2. Zhicheng Sun
3. Pan Li
4. Shaogang Gong
5. Guile Wu

Citations in the corpus

• Class-Incremental Learning for Action Recognition in Videos (2021, neutral): “Although class-incremental learning has been studied for image classification, the research is also active for other applications, including person re-identification [42], 3D object classification [5], object detection [33], and semantic segmentation [24].”
• Patch-based Knowledge Distillation for Lifelong Person Re-Identification (2022, methodology): “[63] characterize lifelong person ReID by unseen class recognition, domain generalization and class imbalanced learning, and then propose a comprehensive learning objective to address these problems.”
• Striking a Balance between Stability and Plasticity for Class-Incremental Learning (2021, methodology): “To transfer knowledge from an old model to a new model, we constrain the distance between normalized embeddings [19, 39] from the new model ( φ ( x ) n ) and the old model ( φ ( x ) o ) as the embedding supervision loss L em , i.e ., Therefore, the overall training objective L can be defined as L =…”
• Learning Continual Compatible Representation for Re-indexing Free Lifelong Person Re-identification (2024, methodology): “…CUHK-SYSU DukeMTMC-ReID MSMT17-V2 CUHK03 Average mAP R@1 mAP R@1 mAP R@1 mAP R@1 mAP R@1 mAP R@1 same pre-processing and evaluation of GwFReID [35], which crops the person images with the ground-truth person bounding box annotation and re-organized as a ReID dataset with corresponding…”