Awesome-Generalized-Category-Discovery

Generalized Category Discovery (GCD) is a advanced task in image recognition that addresses the challenge of categorizing unlabeled images, which may belong to known or novel classes. Unlike traditional methods, GCD does not assume that unlabeled data only comes from known classes or that the number of unknown classes is predetermined.

Contrastive Learning

Generalized Category Discovery (CVPR2022) [paper] [codes]

Guided Cluster Aggregation: A Hierarchical Approach to Generalized Category Discovery (WACV2024) [paper] [codes]

Contrastive Mean-Shift Learning for Generalized Category Discovery (CVPR2024) [paper] [codes]

Self-Distillation & Self-Learning

Parametric Classification for Generalized Category Discovery: A Baseline Study (ICCV2023) [paper] [codes]

No Representation Rules Them All in Category Discovery (NeurIPS2023) [paper] [dataset]

Flipped Classroom: Aligning Teacher Attention with Student in Generalized Category Discovery (NeurIPS2024 Oral) [paper]

Clustering

Parametric Information Maximization for Generalized Category Discovery (ICCV2023) [paper] [codes]

Learning Semi-supervised Gaussian Mixture Models for Generalized Category Discovery (ICCV2023) [paper] [codes]

AMEND: Adaptive Margin and Expanded Neighborhood for Efficient Generalized Category Discovery (WACV2024) [paper]

Prompting

PromptCAL: Contrastive Affinity Learning via Auxiliary Prompts for Generalized Novel Category Discovery (CVPR2023) [paper] [codes]

SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning (ICLR2024) [paper] [codes]

Continual

MetaGCD: Learning to Continually Learn in Generalized Category Discovery (ICCV2023) [paper] [codes]

PromptCCD: Learning Gaussian Mixture Prompt Pool for Continual Category Discovery (ECCV 2024) [paper] [codes]

Happy: A Debiased Learning Framework for Continual Generalized Category Discovery (NeurIPS 2024) [paper] [codes]

Categorization & Hierarchy

Learn to Categorize or Categorize to Learn? Self-Coding for Generalized Category Discovery (NeurIPS2023) [paper] [codes]

Discover and Align Taxonomic Context Priors for Open-world Semi-Supervised Learning (NeruIPS 2023) [paper] [codes]

SelEx: Self-Expertise in Fine-Grained Generalized Category Discovery (ECCV2024) [[paper]](SelEx: Self-Expertise in Fine-Grained Generalized Category Discovery) [codes]

MultiModal

Textual Knowledge Matters: Cross-Modality Co-Teaching for Generalized Visual Class Discovery (ECCV2024) [paper] [codes]