RF-DETR Large model excels at small pollinator detection in video

Researchers have conducted an empirical study on detecting small pollinators in cluttered field videos, comparing YOLO and RF-DETR models. The RF-DETR Large model, when run at a 1344-pixel resolution, achieved the best performance with a mAP50:95 score of 0.405. This result surpassed both a lower-resolution RF-DETR model and the best single-model YOLO baseline. The study found that detector choice and input resolution were more impactful than increased inference complexity, with resolution gains particularly benefiting small objects and rarer classes like bumblebees and moths. AI

IMPACT Improves object detection accuracy for small, occluded targets in complex visual environments.

RANK_REASON Academic paper detailing empirical study and model performance on a specific computer vision task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• alphaXiv
• Bombus
• BuzzSpot
• CatalyzeX
• DagsHub
• Gotit.pub
• Heterocera
• Hugging Face
• Influence Flower
• RF-DETR
• RF-DETR Large
• ScienceCast
• YOLO
• YOLO26m

• paper
• other

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