CSCNET: Class-Specified Cascaded Network for Compositional Zero-Shot Learning (2403.05924v2)
Abstract: Attribute and object (A-O) disentanglement is a fundamental and critical problem for Compositional Zero-shot Learning (CZSL), whose aim is to recognize novel A-O compositions based on foregone knowledge. Existing methods based on disentangled representation learning lose sight of the contextual dependency between the A-O primitive pairs. Inspired by this, we propose a novel A-O disentangled framework for CZSL, namely Class-specified Cascaded Network (CSCNet). The key insight is to firstly classify one primitive and then specifies the predicted class as a priori for guiding another primitive recognition in a cascaded fashion. To this end, CSCNet constructs Attribute-to-Object and Object-to-Attribute cascaded branches, in addition to a composition branch modeling the two primitives as a whole. Notably, we devise a parametric classifier (ParamCls) to improve the matching between visual and semantic embeddings. By improving the A-O disentanglement, our framework achieves superior results than previous competitive methods.
- “From red wine to red tomato: Composition with context,” in Proc. of the CVPR, 2017, pp. 1160–1169.
- “Adversarial fine-grained composition learning for unseen attribute-object recognition,” in Proc. of the ICCV, 2019, pp. 3740–3748.
- “A causal view of compositional zero-shot recognition,” in the NeurIPS, 2020.
- “Dual-stream contrastive learning for compositional zero-shot recognition,” IEEE Transactions on Multimedia, 2023.
- “Learning invariant visual representations for compositional zero-shot learning,” in Proc. of the ECCV, 2022, pp. 339–355.
- “Disentangling visual embeddings for attributes and objects,” in Proc. of the CVPR, 2022, pp. 13658–13667.
- “A decomposable causal view of compositional zero-shot learning,” IEEE Transactions on Multimedia, pp. 1–11, 2022.
- “Relation-aware compositional zero-shot learning for attribute-object pair recognition,” IEEE Transactions on Multimedia, vol. 24, pp. 3652–3664, 2022.
- “Isolating features of object and its state for compositional zero-shot learning,” IEEE Transactions on Emerging Topics in Computational Intelligence, pp. 1–13, 2023.
- “Learning conditional attributes for compositional zero-shot learning,” in Proc. of the CVPR, 2023, pp. 11197–11206.
- “Efficient estimation of word representations in vector space,” in Proc. of the ICLR, 2013.
- “Attributes as operators: Factorizing unseen attribute-object compositions,” in Proc. of the ECCV, 2018, pp. 172–190.
- “Task-driven modular networks for zero-shot compositional learning,” in Proc. of the ICCV, 2019, pp. 3592–3601.
- “Symmetry and group in attribute-object compositions,” in Proc. of the CVPR, 2020, pp. 11313–11322.
- “Open world compositional zero-shot learning,” in Proc. of the CVPR, 2021, pp. 5222–5230.
- “Learning graph embeddings for compositional zero-shot learning,” in Proc. of the CVPR, 2021, pp. 953–962.
- “Swap-reconstruction autoencoder for compositional zero-shot learning,” in Proc. of the ICME, 2023, pp. 438–443.
- “Siamese contrastive embedding network for compositional zero-shot learning,” in Proc. of the CVPR, 2022, pp. 9326–9335.
- “Discovering states and transformations in image collections,” in Proc. of the CVPR, 2015, pp. 1383–1391.
- “Learning attention as disentangler for compositional zero-shot learning,” in Proc. of the CVPR, 2023, pp. 15315–15324.
- “Deep residual learning for image recognition,” in Proc. of the CVPR, 2016, pp. 770–778.
- “Imagenet: A large-scale hierarchical image database,” in Proc. of the CVPR, 2009, pp. 248–255.
- “Enriching word vectors with subword information,” Trans. Assoc. Comput. Linguistics, vol. 5, pp. 135–146, 2017.
- “Adam: A method for stochastic optimization,” in Proc. of the ICLR, 2015.
- “Meta reconciliation normalization for lifelong person re-identification,” in ACM MM, 2022, pp. 541–549.
- “A memorizing and generalizing framework for lifelong person re-identification,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 11, pp. 13567–13585, 2023.
- “Lifelong person re-identification via adaptive knowledge accumulation,” in Proc. of the CVPR, 2021, pp. 7901–7910.
- “Dual gaussian-based variational subspace disentanglement for visible-infrared person re-identification,” in ACM MM, 2020, pp. 2149–2158.
Sponsor
Paper Prompts
Sign up for free to create and run prompts on this paper using GPT-5.
Top Community Prompts
Collections
Sign up for free to add this paper to one or more collections.