TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK
Real-Time Object Detection	COCO	YOLOv4-CSP-P7	MAP	55.4	# 1
Real-Time Object Detection	COCO	YOLOv4-CSP-P7	FPS	16	# 16
Real-Time Object Detection	COCO	YOLOv4-tiny 416	MAP	21.7	# 20
Real-Time Object Detection	COCO	YOLOv4-tiny 416	FPS	371	# 1
Real-Time Object Detection	COCO	YOLOv4-CSP CD53s 640	MAP	47.5	# 4
Real-Time Object Detection	COCO	YOLOv4-CSP CD53s 640	FPS	73	# 3
Object Detection	COCO test-dev	YOLOv4-P5 (CSP-P5, single-scale)	box AP	51.4	# 15
Object Detection	COCO test-dev	YOLOv4-P5 (CSP-P5, single-scale)	AP50	69.9	# 17
Object Detection	COCO test-dev	YOLOv4-P5 (CSP-P5, single-scale)	AP75	56.3	# 18
Object Detection	COCO test-dev	YOLOv4-P5 (CSP-P5, single-scale)	APS	33.1	# 20
Object Detection	COCO test-dev	YOLOv4-P5 (CSP-P5, single-scale)	APM	55.4	# 15
Object Detection	COCO test-dev	YOLOv4-P5 (CSP-P5, single-scale)	APL	62.4	# 19
Object Detection	COCO test-dev	YOLOv4 (CD53)	box AP	45.5	# 44
Object Detection	COCO test-dev	YOLOv4 (CD53)	AP50	64.1	# 49
Object Detection	COCO test-dev	YOLOv4 (CD53)	AP75	49.5	# 47
Object Detection	COCO test-dev	YOLOv4 (CD53)	APS	27	# 45
Object Detection	COCO test-dev	YOLOv4 (CD53)	APM	49	# 43
Object Detection	COCO test-dev	YOLOv4 (CD53)	APL	56.7	# 47
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, single-scale)	box AP	55.4	# 4
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, single-scale)	AP50	73.3	# 4
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, single-scale)	AP75	60.7	# 3
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, single-scale)	APS	38.1	# 2
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, single-scale)	APM	59.5	# 2
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, single-scale)	APL	67.4	# 4
Object Detection	COCO test-dev	YOLOv4-P6 (CSP-P6, single-scale)	box AP	54.3	# 8
Object Detection	COCO test-dev	YOLOv4-P6 (CSP-P6, single-scale)	AP50	72.3	# 8
Object Detection	COCO test-dev	YOLOv4-P6 (CSP-P6, single-scale)	AP75	59.5	# 7
Object Detection	COCO test-dev	YOLOv4-P6 (CSP-P6, single-scale)	APS	36.6	# 6
Object Detection	COCO test-dev	YOLOv4-P6 (CSP-P6, single-scale)	APM	58.2	# 5
Object Detection	COCO test-dev	YOLOv4-P6 (CSP-P6, single-scale)	APL	65.5	# 11
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, multi-scale)	box AP	55.8	# 2
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, multi-scale)	AP50	73.2	# 5
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, multi-scale)	AP75	61.2	# 1
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, multi-scale)	APS	38.8	# 1
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, multi-scale)	APM	60.1	# 1
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, multi-scale)	APL	68.2	# 1

Badge	Markdown
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/scaled-yolov4-scaling-cross-stage-partial/real-time-object-detection-on-coco)](https://paperswithcode.com/sota/real-time-object-detection-on-coco?p=scaled-yolov4-scaling-cross-stage-partial)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/scaled-yolov4-scaling-cross-stage-partial/object-detection-on-coco)](https://paperswithcode.com/sota/object-detection-on-coco?p=scaled-yolov4-scaling-cross-stage-partial)`

Scaled-YOLOv4: Scaling Cross Stage Partial Network

16 Nov 2020 • Chien-Yao Wang • Alexey Bochkovskiy • Hong-Yuan Mark Liao

We show that the YOLOv4 object detection neural network based on the CSP approach, scales both up and down and is applicable to small and large networks while maintaining optimal speed and accuracy. We propose a network scaling approach that modifies not only the depth, width, resolution, but also structure of the network... (read more)

PDF Abstract

Code

Add Remove Mark official

WongKinYiu/ScaledYOLOv4 official

943

AlexeyAB/darknet

↳ Quickstart in

Colab

14,729

Eric3911/ScaledYOLOv4

achen353/Taiwan-Traffic-Object-Detection

frazsaeed/VehicleDetection

↳ Quickstart in

Colab

See all 8 implementations

Tasks

Add Remove

OBJECT DETECTION

REAL-TIME OBJECT DETECTION

Results from the Paper

Edit

Ranked #1 on Real-Time Object Detection on COCO

Get a GitHub badge

TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK	BENCHMARK
Real-Time Object Detection	COCO	YOLOv4-CSP-P7	MAP	55.4	# 1	Compare
Real-Time Object Detection	COCO	YOLOv4-CSP-P7	FPS	16	# 16	Compare
Real-Time Object Detection	COCO	YOLOv4-tiny 416	MAP	21.7	# 20	Compare
Real-Time Object Detection	COCO	YOLOv4-tiny 416	FPS	371	# 1	Compare
Real-Time Object Detection	COCO	YOLOv4-CSP CD53s 640	MAP	47.5	# 4	Compare
Real-Time Object Detection	COCO	YOLOv4-CSP CD53s 640	FPS	73	# 3	Compare
Object Detection	COCO test-dev	YOLOv4-P5 (CSP-P5, single-scale)	box AP	51.4	# 15	Compare
			AP50	69.9	# 17	Compare
			AP75	56.3	# 18	Compare
			APS	33.1	# 20	Compare
			APM	55.4	# 15	Compare
			APL	62.4	# 19	Compare
Object Detection	COCO test-dev	YOLOv4 (CD53)	box AP	45.5	# 44	Compare
			AP50	64.1	# 49	Compare
			AP75	49.5	# 47	Compare
			APS	27	# 45	Compare
			APM	49	# 43	Compare
			APL	56.7	# 47	Compare
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, single-scale)	box AP	55.4	# 4	Compare
			AP50	73.3	# 4	Compare
			AP75	60.7	# 3	Compare
			APS	38.1	# 2	Compare
			APM	59.5	# 2	Compare
			APL	67.4	# 4	Compare
Object Detection	COCO test-dev	YOLOv4-P6 (CSP-P6, single-scale)	box AP	54.3	# 8	Compare
			AP50	72.3	# 8	Compare
			AP75	59.5	# 7	Compare
			APS	36.6	# 6	Compare
			APM	58.2	# 5	Compare
			APL	65.5	# 11	Compare
Object Detection	COCO test-dev	YOLOv4-P7 (CSP-P7, multi-scale)	box AP	55.8	# 2	Compare
			AP50	73.2	# 5	Compare
			AP75	61.2	# 1	Compare
			APS	38.8	# 1	Compare
			APM	60.1	# 1	Compare
			APL	68.2	# 1	Compare

Methods used in the Paper

Add Remove

METHOD	TYPE
Global Average Pooling	Pooling Operations
Batch Normalization	Normalization
Logistic Regression	Generalized Linear Models
Softmax	Output Functions
Residual Connection	Skip Connections
Darknet-53	Convolutional Neural Networks
k-Means Clustering	Clustering
Convolution	Convolutions
Average Pooling	Pooling Operations
Sigmoid Activation	Activation Functions
ReLU	Activation Functions
1x1 Convolution	Convolutions
Max Pooling	Pooling Operations
YOLOv3	Object Detection Models
CSPDarknet53	Convolutional Neural Networks
Spatial Attention Module	Image Model Blocks
Bottom-up Path Augmentation	Feature Extractors
FPN	Feature Extractors
Spatial Pyramid Pooling	Pooling Operations
Label Smoothing	Regularization
CutMix	Image Data Augmentation
Softplus	Activation Functions
Tanh Activation	Activation Functions
DropBlock	Regularization
PAFPN	Feature Extractors
Mish	Activation Functions
Cosine Annealing	Learning Rate Schedules
YOLOv4	Object Detection Models

Scaled-YOLOv4: Scaling Cross Stage Partial Network

Code Edit Add Remove Mark official

Tasks Edit Add Remove

Results from the Paper Edit

Methods used in the Paper Edit Add Remove

Code

Add Remove Mark official

Tasks

Add Remove

Results from the Paper

Edit

Methods used in the Paper

Add Remove