You Only Look Once: Unified, Real-Time Object Detection

Authors: Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. Year: 2015 Venue: CVPR ArXiv: https://arxiv.org/abs/1506.02640

Summary

YOLO introduces single-stage object detection, treating detection as a regression problem. It achieves real-time performance on video and becomes foundational for efficient detection.

Key Concepts

• Single-stage detection pipeline
• Real-time inference (45 FPS)
• Grid-based localization
• Unified detection framework
• Speed-accuracy tradeoff

Impact

YOLO proved single-stage detectors could work and were faster than two-stage methods. It inspired SSD, RetinaNet, and modern efficient detectors.

Related Papers

• SSD: Single Shot MultiBox Detector
• R-CNN: Rich Feature Hierarchies
• RetinaNet: Focal Loss for Dense Object Detection

---

Citation:

@inproceedings{redmon2015you,
  title={You only look once: Unified, real-time object detection},
  author={Redmon, Joseph and Divvala, Santosh and Girshick, Ross and Farhadi, Ali},
  booktitle={IEEE Conference on Computer Vision and Pattern Recognition},
  year={2015}
}