SPECIAL session 3

Chair:
Chen Li, North China University of Technology, China
Email: lichen@ncut.edu.cn

Submission Link: http://www.easychair.org/conferences/?conf=icivc2026 (Select Track Special Session 4)

Introduction:

This session focuses on Few-shot/Zero-shot Visual Learning & Generalization, addressing high-performance visual task challenges under low-data or unseen-category scenarios. It covers key topics: rare-object classification, cross-domain recognition, meta-learning modeling, low-data feature mining, unseen-category detection, semantic segmentation, knowledge distillation, open-set learning, cross-modal transfer and long-tailed data adaptation.
We invite papers on innovative algorithms and transfer strategies. Discussions center on breaking data limitations, with submissions of novel frameworks, case studies and practical solutions welcome.

Topics:

Few-shot visual classification for rare objects
Zero-shot visual recognition across domains
Meta-learning for few-shot visual modeling
Low-data visual feature enhancement & mining
Zero-shot visual object detection for unseen categories
Few-shot semantic segmentation with cross-task transfer
Knowledge distillation for few-shot visual generalization
Open-set zero-shot visual learning for unknown classes
Cross-modal transfer for few-shot visual recognition
Few-shot visual adaptation for long-tailed distribution data