Data to 7 Oct 2026

Semantically Guided Dynamic Visual Prototype Refinement for Compositional Zero-Shot Learning

arXiv 2501.07114 ↗First posted 13 Jan 2025cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Compositional Zero-Shot Learning (CZSL) seeks to recognize unseen state-object pairs by recombining primitives learned from seen compositions. Despite recent progress with vision-language models (VLMs), two limitations remain: (i) text-driven semantic prototypes are weakly discriminative in the visual feature space; and (ii) unseen pairs are optimized passively, thereby inducing seen bias. To address these limitations, we present Duplex, a framework that couples dual-prototype learning with dynamic local-graph refinement of visual prototypes. For each composition, Duplex maintains a semantic prototype via prompt learning and a visual prototype for unseen pairs constructed by recombining disentangled state and object primitives from seen images. The visual prototypes are updated dynamically through lightweight aggregation on mini-batch local graphs, which incorporates unseen compositions during training without labels. This design introduces fine-grained visual evidence while preserving semantic structure. It enriches class prototypes, better disambiguates semantically similar yet visually distinct pairs, and mitigates seen bias. Experiments on MIT-States, UT-Zappos, and CGQA in closed-world and open-world settings achieve competitive performance and consistent compositional generalization. Our source code is available at https://github.com/ISPZ/Duplex-CZSL.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 42 of its 63 references that are in the map. First posted on arXiv 13 Jan 2025.

Builds on

42 of 63 references.

References · 42

2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
2020Language Models are Few-Shot Learnerscited by 32,199
2016Semi-Supervised Classification with Graph Convolutional Networkscited by 11,282
2021Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervisioncited by 3,848
2017Prototypical Networks for Few-shot Learningcited by 3,001
2021Learning to Prompt for Vision-Language Modelscited by 2,269
2021ClipCap: CLIP Prefix for Image Captioningcited by 557
2021How Much Can CLIP Benefit Vision-and-Language Tasks?cited by 321
2021An Empirical Study of GPT-3 for Few-Shot Knowledge-Based VQAcited by 298
2019Task-Driven Modular Networks for Zero-Shot Compositional Learningcited by 121
2021Learning Graph Embeddings for Compositional Zero-shot Learningcited by 112
2021Open World Compositional Zero-Shot Learningcited by 100

30 more

2020Symmetry and Group in Attribute-Object Compositionscited by 89
2020A causal view of compositional zero-shot recognitioncited by 78
2023GraphAdapter: Tuning Vision-Language Models With Dual Knowledge Graphcited by 69
2022Learning to Compose Soft Prompts for Compositional Zero-Shot Learningcited by 66
2022Disentangling Visual Embeddings for Attributes and Objectscited by 55
2022Siamese Contrastive Embedding Network for Compositional Zero-Shot Learningcited by 55
2023CLIP-Guided Vision-Language Pre-training for Question Answering in 3D Scenescited by 55
2021Learning Graph Embeddings for Open World Compositional Zero-Shot Learningcited by 49
2021Independent Prototype Propagation for Zero-Shot Compositionalitycited by 42
2022KG-SP: Knowledge Guided Simple Primitives for Open World Compositional Zero-Shot Learningcited by 41
2022Decomposed Soft Prompt Guided Fusion Enhancing for Compositional Zero-Shot Learningcited by 36
2023ConZIC: Controllable Zero-shot Image Captioning by Sampling-Based Polishingcited by 33
2020SumGraph: Video Summarization via Recursive Graph Modelingcited by 32
2023Transductive Few-shot Learning with Prototype-based Label Propagation by Iterative Graph Refinementcited by 30
2023Learning Attention as Disentangler for Compositional Zero-shot Learningcited by 29
2023Learning Conditional Attributes for Compositional Zero-Shot Learningcited by 28
2021Dual Progressive Prototype Network for Generalized Zero-Shot Learningcited by 26
2023Troika: Multi-Path Cross-Modal Traction for Compositional Zero-Shot Learningcited by 26
2022Learning Invariant Visual Representations for Compositional Zero-Shot Learningcited by 24
2022Learning Attention Propagation for Compositional Zero-Shot Learningcited by 20
2024Context-based and Diversity-driven Specificity in Compositional Zero-Shot Learningcited by 20
2023Hierarchical Visual Primitive Experts for Compositional Zero-Shot Learningcited by 16
2023GIPCOL: Graph-Injected Soft Prompting for Compositional Zero-Shot Learningcited by 16
2024Visual-Augmented Dynamic Semantic Prototype for Generative Zero-Shot Learningcited by 16
2021A Closer Look at Prototype Classifier for Few-shot Image Classificationcited by 15
2022On Leveraging Variational Graph Embeddings for Open World Compositional Zero-Shot Learningcited by 13
2022Prompting Large Pre-trained Vision-Language Models For Compositional Concept Learningcited by 12
2023Patch-Prompt Aligned Bayesian Prompt Tuning for Vision-Language Modelscited by 12
2023Distilled Reverse Attention Network for Open-world Compositional Zero-Shot Learningcited by 11
2022ProCC: Progressive Cross-primitive Compatibility for Open-World Compositional Zero-Shot Learningcited by 9

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way