Zero-shot Generalizable Incremental Learning for Vision-Language Object Detection

arXiv:2403.01680v1 Announce Type: new
Abstract: This paper presents Incremental Vision-Language Object Detection (IVLOD), a novel learning task designed to incrementally adapt pre-trained Vision-Language Object Detection Models (VLODMs) to various specialized domains, while simultaneously preserving their zero-shot generalization capabilities for the generalized domain. To address this new challenge, we present the Zero-interference Reparameterizable Adaptation (ZiRa), a novel method that introduces Zero-interference Loss and reparameterization techniques to tackle IVLOD without incurring additional inference costs or a significant increase in memory usage. Comprehensive …

abstract adapt arxiv capabilities challenge cs.cv detection domain domains generalized incremental interference language novel paper type vision zero-shot