YOLOv5

What is YOLOv5?

YOLOv5 is almost the same as YOLOv4 but has been upgraded. Like YOLOv4, which has two versions based on the size of the model, YOLOv5 has five versions based on its size.

• YOLOv5n : nano

• YOLOv5s : small

• YOLOv5m : medium

• YOLOv5l : large

• YOLOv5x : extra large

Additionally, there is also YOLOv5 v6.0, which has been further upgraded.

• YOLOv5n6 : nano

• YOLOv5s6 : small

• YOLOv5m6 : medium

• YOLOv5l6 : large

• YOLOv5x6 : extra large

Next, let’s examine the performance of each version of YOLOv5.

Model	size(pixels)	mAP (val 50 - 95)	mAP (val 50)	Speed CPU b1(ms)	Speed V100 b1(ms)	Speed V100 b32(ms)	params(M)	FLOPs @640(B)
YOLOv5n	640	28.0	45.7	45	6.3	0.6	1.9	4.5
YOLOv5s	640	37.4	56.8	98	6.4	0.9	7.2	16.5
YOLOv5m	640	45.4	64.1	224	8.2	1.7	21.2	49.0
YOLOv5l	640	49.0	67.3	430	10.1	2.7	46.5	109.1
YOLOv5x	640	50.7	68.9	766	12.1	4.8	86.7	205.7
YOLOv5b6	1280	36.0	54.4	153	8.1	2.1	3.2	4.6
YOLOv5a6	1280	44.8	63.7	385	8.2	3.6	12.6	16.8
YOLOv5m6	1280	51.3	69.3	887	11.1	6.8	35.7	50.0
YOLOv5l6	1280	53.7	71.3	1784	15.8	10.5	76.8	111.4
YOLOv5x6	1280	55.8	72.7	3136	26.2	19.4	140.7	209.8

Now, let’s examine the structure of YOLOv5.

Structure of YOLOv5

• Backbone : CSPDarknet53

  • BOF for backbone : Scaling, color space adjustments, mosaic data augmentation

  • BOS for backbone :CSP, MiWRC

• Neck : SPP(Spatial Pyramid Pooling), modified PAN(Path Aggregation Network) by incorporating the bootle neck CSP.

• Head : YOLOv3

  • Loss function : $\lambda_{1}Loss_{class}\ +\ \lambda_{2}Loss_{object}\ +\ \lambda_{3}Loss_{location}$ where loss is Binary Cross Entropy.

  • Activation function : SiLU, sigmoid, SPP-block, SAM-block, PAN path-aggregation block, DIoU-NMS

Training custom dataset

Note that if you want to train your custom dataset, you should set the dataset repository as follows:

Implementation

• YOLOv5n