[Attention] Which is your most desired variant of the YOLO series?

[RangeKing]

MMYOLO Call for Proposals

在各个领域中 YOLO 系列改进算法众多，哪些改进的算法或者模块你觉得非常有用？欢迎在 Issue 中留言。我们会根据反馈列入开发计划当中。

Issue 回复模板

• 算法名：
• 发布的会议和年份（如有）：
• 论文（如有）：
• 数据集（如有）：
• 简介（如有）：
• 代码（如有）：
• 相关解读（如有）：

示例

• 算法名：TPH-YOLOv5
• 发布的会议和年份：ICCVW2021
• 论文：https://openaccess.thecvf.com/content/ICCV2021W/VisDrone/html/Zhu_TPH-YOLOv5_Improved_YOLOv5_Based_on_Transformer_Prediction_Head_for_Object_ICCVW_2021_paper.html
• 数据集：VisDrone2021
• 简介：YOLOv5 基础上增加 Transformer Prediction Heads (TPH) 和 convolutional block attention model (CBAM)，在 VisDrone2021 数据集上比 baseline YOLOv5 模型，精度提高 7%
• 代码：https://github.com/cv516Buaa/tph-yolov5
• 相关解读：https://zhuanlan.zhihu.com/p/416171744

---

There are lots of improved algorithms in the YOLO series in various fields. Which improved algorithms or modules do you find useful? Feel free to leave your comments in the Issue. We will include them in the development plan based on the feedback.

Issue Reply Template

• Algorithm Name:
• Conference/Year(Optional):
• Paper(Optional):
• Datasets(Optional):
• Introduction(Optional):
• Repo(Optional):
• Interpretations(Optional):

Example

• Algorithm Name: TPH-YOLOv5
• Conference/Year: ICCVW2021
• Paper: https://openaccess.thecvf.com/content/ICCV2021W/VisDrone/html/Zhu_TPH-YOLOv5_Improved_YOLOv5_Based_on_Transformer_Prediction_Head_for_Object_ICCVW_2021_paper.html
• Datasets: VisDrone2021
• Introduction: Add Transformer Prediction Heads (TPH) and convolutional block attention model (CBAM) in YOLOv5, which improves the accuracy by 7% over the baseline YOLOv5 model on the VisDrone2021 dataset.
• Repo: https://github.com/cv516Buaa/tph-yolov5
• Interpretations: https://zhuanlan.zhihu.com/p/416171744

---

根据大家反馈，我们会在下面维护一个列表，方便统一查看。
According to your feedback, we will maintain a list below.

Algorithm Name	Conference/Year	Paper	Datasets	Introduction	Repo	Interpretations
TPH-YOLOv5	ICCVW2021	link	VisDrone2021	TPH/CBAM	Official	Zhihu1
YOLOV	2022	link	IMAGENET VID and DET		Official
Loss: EFL	2022	link	LVIS v1	EFL	Official	Zhihu1
CFPNet	2022	link	MS COCO	Centralized Feature Pyramid for Object Detection	Official	mmyolo
YOLOP	2022	link	BDD100K	You Only Look at Once for Panoptic driving Perception	Official	Zhihu1
YOLO-Pose	2022	link	COCO Keypoint	Enhancing YOLO for Multi Person Pose Estimation Using Object Keypoint Similarity Loss	Official	Zhihu1
Edge-YOLO	2023	link	COCO and Visdrone	EdgeYOLO: An Edge-Real-Time Object Detector	Official
YOLO-NAS	2023		MS COCO	A Next-Generation, Object Detection Foundational Model generated by Deci’s Neural Architecture Search Technology	Official	Zhihu1
YOLO-MS	2023	link	MS COCO	Rethinking Multi-Scale Representation Learning for Real-time Object Detectionblock	Official	Weixin

[xin-li-67]

Algorithm Name： YOLOV
Conference/Year：2022
Paper：link
Datasets：ILSVRC2015 DET and ILSVRC2015 VID dataset on IMAGENET
Introduction：N/A
Repo：https://github.com/YuHengsss/YOLOV
Interpretations：N/A

[SicongLiu998]

Algorithm Name：Equalized Focal Loss for Dense Long-Tailed Object Detection
Conference/Year(Optional)：CVPR 2022
Paper(Optional)：https://arxiv.org/pdf/2201.02593.pdf
Datasets(Optional)：LVIS v1 dataset
Introduction(Optional)：N/A
Repo(Optional)：https://github.com/ModelTC/United-Perception
Interpretations(Optional)：https://zhuanlan.zhihu.com/p/489606679

[hhaAndroid]

Algorithm Name： Centralized Feature Pyramid for Object Detection
Conference/Year：2022
Paper：link
Datasets：MS COCO
Introduction：Centralized Feature Pyramid for Object Detection
Repo：https://github.com/QY1994-0919/CFPNet
Interpretations：QY1994-0919/CFPNet#10

[debapriyamaji]

Algorithm Name： YOLOP: You Only Look at Once for Panoptic driving Perception
Conference/Year：2022
Paper：link
Datasets: BDD100K
Introduction：
Repo：https://github.com/hustvl/YOLOP
Interpretations：

[debapriyamaji]

Algorithm Name："YOLO-Pose: Enhancing YOLO for Multi Person Pose Estimation Using Object Keypoint Similarity Loss]
Conference/Year：2022
Paper：link
Datasets: COCO Keypoint
Introduction：
Repo：https://github.com/TexasInstruments/edgeai-yolov5/tree/yolo-pose
Interpretations：

[RangeKing]

Algorithm Name："YOLO-Pose: Enhancing YOLO for Multi Person Pose Estimation Using Object Keypoint Similarity Loss] Conference/Year：2022 Paper：link Datasets: COCO Keypoint Introduction： Repo：https://github.com/TexasInstruments/edgeai-yolov5/tree/yolo-pose Interpretations：

Hi @debapriyamaji, thanks for your attention to MMYOLO. I think YOLO-Pose is an excellent algorithm. Could you help MMYOLO support YOLO-Pose. We can open another issue to discuss it.

[debapriyamaji]

Hi @RangeKing ,

Thanks for the reply. I would like to enable yolo-pose in mmYOLO with your help. Right now, I am not sure when can I spend some time on it. I will get back to you with some updates within the next couple of weeks.

Regards, Debapriya

[zcunyi]

Hi @RangeKing ,

Thanks for the reply. I would like to enable yolo-pose in mmYOLO with your help. Right now, I am not sure when can I spend some time on it. I will get back to you with some updates within the next couple of weeks.

Regards, Debapriya
Hi, @debapriyamaji , I am using yolo-pose, looking forward to your work.

[wang-tf]

Algorithm Name：YOLO-NAS
Conference/Year：2023
Datasets：MS COCO
Introduction：providing superior real-time object detection capabilities and production-ready performance
Repo：https://github.com/Deci-AI/super-gradients/blob/master/YOLONAS.md
Interpretations：https://zhuanlan.zhihu.com/p/628090349