【Yolov11】《Yolov11: An overview of the key architectural enhancements》

arXiv-2024

代码：

• https://github.com/ultralytics/ultralytics

文档：

• Ultralytics YOLO11
• 使用 Ultralytics YOLO 进行模型训练

---

---

1、Background and Motivation

Background

• 计算机视觉的快速发展
• YOLO系列模型的演进
• YOLOv11的发布

Motivation

• 提升对象检测性能
• 优化模型效率和可扩展性
• 推动实时计算机视觉应用的发展

This study presents an architectural analysis of YOLOv11, the latest iteration in the YOLO (You Only Look Once) series of object detection models.

芒果YOLO11算法解析-最新YOLO11结构图以及YOLO11各部分细致结构图解析

2、Related Work

yolov1 ~ yolov10

3、Evolution of YOLO models

这个 contributions 不知道准不准确，早期的 yolov5 应该是 anchor-based ，不过引用[10] 指向的不是 v5 官网地址

4、Architectural footprint of Yolov11

核心改动是 C3K2（ Cross Stage Partial with kernel size 2） 和 C2PSA (Convolutional block with Parallel Spatial Attention)

看源码的描述 2 好像是 two convolution 的意思 （two c3k），并非 kernel size

# YOLO11n backbone
backbone:
  # [from, repeats, module, args]
  - [-1, 1, Conv, [64, 3, 2]] # 0-P1/2
  - [-1, 1, Conv, [128, 3, 2]] # 1-P2/4
  - [-1, 2, C3k2, [256, False, 0.25]]
  - [-1, 1, Conv, [256, 3, 2]] # 3-P3/8
  - [-1, 2, C3k2, [512, False, 0.25]]
  - [-1, 1, Conv, [512, 3, 2]] # 5-P4/16
  - [-1, 2, C3k2, [512, True]]
  - [-1, 1, Conv, [1024, 3, 2]] # 7-P5/32
  - [-1, 2, C3k2, [1024, True]]
  - [-1, 1, SPPF, [1024, 5]] # 9
  - [-1, 2, C2PSA, [1024]] # 10

# YOLO11n head
head:
  - [-1, 1, nn.Upsample, [None, 2, "nearest"]]
  - [[-1, 6], 1, Concat, [1]] # cat backbone P4
  - [-1, 2, C3k2, [512, False]] # 13

  - [-1, 1, nn.Upsample, [None, 2, "nearest"]