UniDGF: A Unified Detection-to-Generation Framework for Hierarchical Object Visual Recognition
This addresses the challenge of capturing fine-grained distinctions and attribute diversity in large-scale e-commerce scenarios, representing an incremental improvement over current approaches.
The paper tackles the problem of unified visual semantic understanding for hierarchical object recognition by introducing a detection-guided generative framework that predicts category and attribute tokens, achieving stronger fine-grained recognition and more coherent unified inference compared to existing similarity-based and multi-stage systems.
Achieving visual semantic understanding requires a unified framework that simultaneously handles object detection, category prediction, and attribute recognition. However, current advanced approaches rely on global similarity and struggle to capture fine-grained category distinctions and category-specific attribute diversity, especially in large-scale e-commerce scenarios. To overcome these challenges, we introduce a detection-guided generative framework that predicts hierarchical category and attribute tokens. For each detected object, we extract refined ROI-level features and employ a BART-based generator to produce semantic tokens in a coarse-to-fine sequence covering category hierarchies and property-value pairs, with support for property-conditioned attribute recognition. Experiments on both large-scale proprietary e-commerce datasets and open-source datasets demonstrate that our approach significantly outperforms existing similarity-based pipelines and multi-stage classification systems, achieving stronger fine-grained recognition and more coherent unified inference.