YOLOv4

Introduced by Bochkovskiy et al. in YOLOv4: Optimal Speed and Accuracy of Object Detection

YOLOv4 is a one-stage object detection model that improves on YOLOv3 with several bags of tricks and modules introduced in the literature. The components section below details the tricks and modules used.

Source: YOLOv4: Optimal Speed and Accuracy of Object Detection

Read Paper See Code

Papers

Paper	Code	Results	Date	Stars

Tasks

Task	Papers	Share
Object Detection	54	26.34%
Object	28	13.66%
Real-Time Object Detection	8	3.90%
Deep Learning	7	3.41%
Autonomous Driving	5	2.44%
Semantic Segmentation	5	2.44%
Domain Adaptation	3	1.46%
Image Classification	3	1.46%
Traffic Sign Detection	3	1.46%

Usage Over Time

This feature is experimental; we are continuously improving our matching algorithm.

Components

Component	Type	Add Remove
Cosine Annealing	Learning Rate Schedules
CSPDarknet53	Convolutional Neural Networks
CutMix	Image Data Augmentation
DropBlock	Regularization
Grid Sensitive	Object Detection Modules
Label Smoothing	Regularization
Mish	Activation Functions
PAFPN	Feature Extractors
Spatial Attention Module	Image Model Blocks
Spatial Pyramid Pooling	Pooling Operations
YOLOv3	Object Detection Models

Categories

Add Remove

Object Detection Models

One-Stage Object Detection Models