Transformer Model Excels at Detecting Tiny Pollinators in ECCV Challenge

Researchers have developed a novel Transformer-based model for detecting tiny pollinators in challenging field keyframes. The model, named Co-DINO with a Swin-L backbone, achieved the highest mean Average Precision (mAP) in comparisons. To address the dominance of bees in the dataset and the small size of the target objects, the training process incorporated a crop-mosaic pool to increase the representation of rarer insect classes and an ETF loss function to better distinguish between classes. This approach led to a first-place ranking on the FinalTest leaderboard for the CVPPA@ECCV 2026 BuzzSpot Challenge. AI

IMPACT This research advances computer vision techniques for detecting small objects, potentially improving ecological monitoring and biodiversity studies.

RANK_REASON The item is an academic paper detailing a new model and methodology for a computer vision task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• alphaXiv
• CatalyzeX Code Finder for Papers
• Co-DINO
• Connected Papers
• CORE Recommender
• CVPPA@ECCV 2026 BuzzSpot Challenge
• DagsHub
• ETF loss
• FinalTest
• Gotit.pub
• Hugging Face
• Influence Flower
• Litmaps
• ScienceCast
• scite Smart Citations
• Swin-L

• paper
• model release
• other

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