List of YOLO Models for Object Detection

Anyone working in the field of deep learning for solving computer vision tasks has heard of YOLO. If this actually is a useful architecture for many real-life deployment cases is something I’ll leave open here. It certainly is useful for some quick initial prototypes if object detection is required. Some comparison of model performance can be found here.

Over the past years an enormous amount of object detectors are called YOLO. It certainly got out of control. Anyhow, here is a list of them:

The darknet family

• YOLO (You Only Look Once)

• YOLO 9000 (YOLOv2)

• YOLOv3
  • the last version of the guy who started it initially
• YOLOv4
  • most likely the most used of the original darknet-based version

Based on PyTorch and other frameworks

• PP-YOLO

• PP-YOLOE

• Scaled YOLOv4

• YOLOv5
  • most likely one of the most popular versions as it allows easy model export in all kind of formats from its trained pytorch models

• YOLOv6

• YOLOv7

• YOLOv8
  • successor of YOLOv5 by the same company

• YOLOv9

• YOLOv10

• YOLOv11
  • successor of YOLOv8 by the same company
• YOLOv12
  • “Attention-Centric Real-Time Object Detectors”
• YOLOv13
  • “Real-Time Object Detection with Hypergraph-Enhanced Adaptive Visual Perception”
  • paper
• YOLOv14
  • “YOLOv14: Adaptive Real-Time Object Detection for Diverse Imaging Conditions”
  • paper
• YOLO26
  • “YOLO26: Key Architectural Enhancements and Performance Benchmarking for Real-Time Object Detection”
  • paper
• YOLO-NAS
  • model generated using neural architecture search

• YOLOR

• YOLO-World
  • “YOLO-World: Real-Time Open-Vocabulary Object Detection”
  • paper

• YOLOX

• DAMO-YOLO

• Gold-YOLO