New DRAFE model enhances traffic object detection across cities

Researchers have developed DRAFE, a novel ensemble method for traffic object detection that improves cross-city generalization and fine-grained recognition. DRAFE combines two independently trained detection transformers, LW-DETR and RF-DETR, using a two-stage training strategy. This approach involves pre-training on a curated corpus of over 6,000 images and then fine-tuning on the Project Hafnia Track 6 dataset. The system achieved a 0.4022 mAP on the AI City Challenge 2026 Track 6, securing sixth place among 25 teams. AI

IMPACT Enhances fine-grained traffic object detection and cross-city generalization, potentially improving intelligent transportation systems.

RANK_REASON The cluster describes a new research paper detailing a novel model for a specific computer vision task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• AI City Challenge 2026 Track 6
• DRAFE
• LW-DETR
• RF-DETR

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →