Anatomy-Guided Vision-Language Learning with Angular Prototype Separation for Multi-Label Video Capsule Endoscopy Classification Under Class Imbalance
📰 ArXiv cs.AI
Learn to improve multi-label video capsule endoscopy classification using anatomy-guided vision-language learning and angular prototype separation under class imbalance
Action Steps
- Implement Angular Separation Loss on class prototypes to address class imbalance
- Utilize a Biological State Machine temporal decoder for temporal event detection
- Fuse consecutive frames using LocaL fusion for improved feature extraction
- Apply BiomedCLIP as the backbone for vision-language learning
- Evaluate the performance of the proposed framework on the Galar dataset
Who Needs to Know This
This research benefits computer vision engineers and biomedical researchers working on medical image analysis and classification tasks, particularly those dealing with class imbalance issues.
Key Insight
💡 Combining angular prototype separation with a biological state machine temporal decoder can effectively address class imbalance in multi-label video capsule endoscopy classification
Share This
📊 Improve multi-label video capsule endoscopy classification with anatomy-guided vision-language learning and angular prototype separation! 🚀
Key Takeaways
Learn to improve multi-label video capsule endoscopy classification using anatomy-guided vision-language learning and angular prototype separation under class imbalance
Full Article
Title: Anatomy-Guided Vision-Language Learning with Angular Prototype Separation for Multi-Label Video Capsule Endoscopy Classification Under Class Imbalance
Abstract:
arXiv:2603.17879v2 Announce Type: replace-cross Abstract: This work presents a multi-label temporal event detection framework for video capsule endoscopy (VCE) that addresses the extreme class imbalance inherent in the Galar dataset by combining two principal contributions: an Angular Separation Loss on class prototypes and a Biological State Machine temporal decoder. The backbone remains BiomedCLIP, a biomedical vision-language foundation model. Three consecutive frames are fused through a Loca
Abstract:
arXiv:2603.17879v2 Announce Type: replace-cross Abstract: This work presents a multi-label temporal event detection framework for video capsule endoscopy (VCE) that addresses the extreme class imbalance inherent in the Galar dataset by combining two principal contributions: an Angular Separation Loss on class prototypes and a Biological State Machine temporal decoder. The backbone remains BiomedCLIP, a biomedical vision-language foundation model. Three consecutive frames are fused through a Loca
DeepCamp AI