UniDGF: A Unified Detection-to-Generation Framework for Hierarchical Object Visual Recognition

2025/11/20 by Nan, Xinyu, Mao, Lingtao, Dai, Huangyu +8
#Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.2511.15984

Abstract

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.

Citations

• OneSearch: A Preliminary Exploration of the Unified End-to-End Generative Framework for E-commerce Search
• MOON: Generative MLLM-based Multimodal Representation Learning for E-commerce Product Understanding
• Qwen2.5-VL Technical Report
• YOLOv11: An Overview of the Key Architectural Enhancements
• BInD: Bond and Interaction-generating Diffusion Model for Multi-objective Structure-based Drug Design
• YOLO-World: Real-Time Open-Vocabulary Object Detection
• Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling
• Vision Transformer with Quadrangle Attention
• Annotated Point Clouds and Images from a Spatial-Semantic Perception Pipeline: Robotized Deconstruction at the Reference Construction Site, Aachen
• OvarNet: Towards Open-vocabulary Object Attribute Recognition
• Transformer Memory as a Differentiable Search Index
• Learning to Predict Visual Attributes in the Wild
• Learning Transferable Visual Models From Natural Language Supervision
• BART: Denoising Sequence-to-Sequence Pre-training for Natural Language\n Generation, Translation, and Comprehension
• Cascade R-CNN: Delving into High Quality Object Detection
• Mask R-CNN
• Microsoft COCO: Common Objects in Context

Related

• Large Scale Generative Multimodal Attribute Extraction for E-commerce Attributes
• SEAL: Semantic-Aware Hierarchical Learning for Generalized Category Discovery
• eProduct: A Million-Scale Visual Search Benchmark to Address Product Recognition Challenges
• Cross-Platform E-Commerce Product Categorization and Recategorization: A Multimodal Hierarchical Classification Approach
• GlideNet: Global, Local and Intrinsic based Dense Embedding NETwork for Multi-category Attributes Prediction