Reproduce “Ultralytics YOLO Evolution: An Overview of YOLO27, YOLO26, YOLO11, YOLOv8, and YOLOv5 Object Detectors for Computer Vision and Pattern Recognition” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Ranjan Sapkota, Manoj Karkee

YOLO27 adapts object-detection architecture to model scale: compact n/s variants use CNN pathways with dense predictions and optional NMS-free inference, while larger m/l variants use learned queries and transformer decoders for direct detections. The paper reports 60.4 COCO mAP50–95 at 640-pixel input and 2.32 ms latency for YOLO27l, but figures remain preliminary pending release and independent evaluation.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.