Awesome-Compositional-Zero-Shot-Learning-Papers

Awesome Compositional Zero-shot Learning papers. Generally, composition zero-shot learning methods are divided into three different classes: disentangled CZSL methods, composed CZSL methods and CLIP-based methods, distinguished by the distribution of attributes and objects. Due to the limit of my comprehension, the following classifications are for reference only, papers are the key points.：）

Paper List

• Disentangled CZSL Methods
• Composed CZSL Methods
• CLIP-based CZSL Methods

Disentangled CZSL Methods

• Simple Primitives With Feasibility- and Contextuality-Dependence for Open-World Compositional Zero-Shot Learning(SAD-SP) (2024 TPAMI), Zhe Liu et al. [pdf]
• Learning Attention as Disentangler for Compositional Zero-shot Learning(ADE) (2023 CVPR), Shaozhe Hao et al. [pdf]
• Leveraging Sub-class Discimination for Compositional Zero-Shot Learning (2023 AAAI), Xiaoming Hu et al. [pdf]
• Disentangling Visual Embeddings for Attributes and Objects(OADis) (2022 CVPR), Nirat Saini et al. [pdf]
• KG-SP: Knowledge Guided Simple Primitives for Open World Compositional Zero-Shot Learning(KG-SP) (2022 CVPR), Shyamgopal Karthik et al. [pdf]
• Learning Invariant Visual Representations for Compositional Zero-Shot Learning (2022 ECCV), Tian Zhang et al. [pdf]
• Siamese Contrastive Embedding Network for Compositional Zero-Shot Learning(SCEN) (2022 CVPR), Xiangyu Li et al. [pdf]
• Independent Prototype Propagation for Zero-Shot Compositionality(ProtoProp) (2021 NeurIPS), Frank Ruis et al. [pdf]
• Learning Single/Multi-Attribute of Object With Symmetry and Group(SymNet) (2021 TPAMI), Yong-Lu Li et al. [pdf]
• Revisiting Visual Product for Compositional Zero-Shot Learning(VisProd) (2021 NeurIPS), Shyamgopal Karthik et al. [pdf]
• Relation-Aware Compositional Zero-Shot Learning for Attribute-Object Pair Recognition(BMP-Net) (2021 TMM), Ziwei Xu et al. [pdf]
• A Causal View of Compositional Zero-shot Recognition (2020 NeurIPS), Yuval Atzmon et al. [pdf]
• Learning Unseen Concepts via Hierarchical Decomposition and Composition(HiDC) (2020 CVPR), Muli Yang et al. [pdf]
• Symmetry and Group in Attribute-Object Compositions(SymNet) (2020 CVPR), Yong-Lu Li et al. [pdf]
• Attributes as Operators: Factorizing Unseen Attribute-Object Compositions(AttrAsOp) (2018 ECCV), Tushar Nagarajan et al. [pdf]
• From Red Wine to Red Tomato: Composition with Context (2017 CVPR), Ishan Misra et al. [pdf]

Composed CZSL Methods

• Learning Conditional Attributes for Compositional Zero-Shot Learning(CANet) (2023 CVPR), Qingsheng Wang et al. [pdf]
• Learning to Infer Unseen Single-Multi-Attribute-Object Compositions With Graph Networks (2023 TPAMI), Hui Chen et al. [pdf]
• On Leveraging Variational Graph Embeddings for Open World Compositional Zero-Shot Learning(CVGAE) (2022 ACM Multimedia), Muhammad Umer Anwaar et al. [pdf]
• Learning Graph Embeddings for Compositional Zero-Shot Learning(CGE) (2021 CVPR), Muhammad Ferjad Naeem et al. [pdf]
• Learning Graph Embeddings for Open World Compositional Zero-Shot Learning(Co-CGE) (2021 TPAMI), Massimiliano Mancini et al. [pdf]
• Open World Compositional Zero-Shot Learning(CompCos) (2021 CVPR), Massimiliano Mancini et al. [pdf]
• Adversarial Fine-Grained Composition Learning for Unseen Attribute-Object Recognition (2019 ICCV), Kun Wei et al. [pdf]
• Task-Driven Modular Networks for Zero-Shot Compositional Learning(TMN) (2019 ICCV), Senthil Purushwalkamet al. [pdf]

CLIP-based CZSL Methods

• CAILA: Concept-Aware Intra-Layer Adapters for Compositional Zero-shot Learning(CAILA) (2024 WACV), Zhaoheng Zheng et al. [pdf]
• GIPCOL: Graph-Injected Soft Prompting for Compositional Zero-Shot Learning(GIPCOL) (2024 WACV), Guangyue Xu et al. [pdf]
• Learning to Compose Soft Prompts for Compositional Zero-Shot Learning(CSP) (2023 ICLR), Nihal V. Nayak et al. [pdf]
• Decomposed Soft Prompt Guided Fusion Enhancing for Compositional Zero-Shot Learning(DFSP) (2023 CVPR), Xiaocheng Lu et al. [pdf]