LMPT: Prompt Tuning with Class-Specific Embedding Loss for Long-tailed Multi-Label Visual Recognition
This addresses the challenging problem of imbalanced data distribution and label co-occurrence in multi-label visual recognition, offering improved performance on both head and tail classes, though it appears incremental as it builds on existing prompt tuning and loss techniques.
The paper tackles long-tailed multi-label visual recognition by proposing LMPT, a prompt tuning framework with class-specific embedding loss, which significantly surpasses previous state-of-the-art methods and zero-shot CLIP on VOC-LT and COCO-LT datasets.
Long-tailed multi-label visual recognition (LTML) task is a highly challenging task due to the label co-occurrence and imbalanced data distribution. In this work, we propose a unified framework for LTML, namely prompt tuning with class-specific embedding loss (LMPT), capturing the semantic feature interactions between categories by combining text and image modality data and improving the performance synchronously on both head and tail classes. Specifically, LMPT introduces the embedding loss function with class-aware soft margin and re-weighting to learn class-specific contexts with the benefit of textual descriptions (captions), which could help establish semantic relationships between classes, especially between the head and tail classes. Furthermore, taking into account the class imbalance, the distribution-balanced loss is adopted as the classification loss function to further improve the performance on the tail classes without compromising head classes. Extensive experiments are conducted on VOC-LT and COCO-LT datasets, which demonstrates that our method significantly surpasses the previous state-of-the-art methods and zero-shot CLIP in LTML. Our codes are fully public at https://github.com/richard-peng-xia/LMPT.