added lighter yolov5 config files

[jere357]

Hello, this PR containts 3 config files for much lighter yolov5 detection models

The smallest yolov5 model on the repo currently is yolov5n and it scores like 37 mAP on COCO. This makes the model look bad when it actually performs really well on real world tasks, The mAP is low because COCO is a very difficult dataset and has 80 classes, while in most real world situations you will usually have only several classes in yout dataset, my personal problem was "roughly detecting shelves" in a store and that dataset was solvable with only femto configuration. I think adding these couple of files here may encourage people to try lighter and faster models because the yolov5n is a deceptively good object detection model, benchmarking it on 80 classes is unrealistic for many real world uses of the yolov object detector :)

Feel free to open a discussion about this in the comments,
maybe casually mention these configs somewhere in the docs

🤖 Generated by Copilot at f134a68

Summary

⚡🚀💡

This pull request adds three new YAML files that define very small and fast YOLOv5 models with different depth and width multipliers. These files are yolov5femto.yaml, yolov5micro.yaml, and yolov5pico.yaml. They are intended to offer more choices for users who need minimal models for resource-constrained applications.

yolov5 models
smaller and faster options
autumn of deep learning

Walkthrough

• Add three new YAML files for very small and fast YOLOv5 models: yolov5femto.yaml, yolov5micro.yaml, and yolov5pico.yaml (link, link, link)

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

📊 Key Changes

• New YOLOv5 configurations named yolov5femto.yaml, yolov5micro.yaml, and yolov5pico.yaml have been added, each with varying depth_multiple and width_multiple values to create lighter models.

🎯 Purpose & Impact

• The addition of these lighter configurations is aimed at enabling the YOLOv5 object detection model to run on devices with limited computational resources, such as edge devices or systems with low processing power. This could greatly expand the applicability of the YOLOv5 model in real-world scenarios where computational efficiency is key.

🌟 Summary

New YOLOv5 configurations introduced for resource-constrained environments, enhancing the model's versatility for edge computing. 🚀

[jere357]

@glenn-jocher any updates regarding these lightweight configs?
#11812 - this PR was the messy code one
this PR here is the one with "nice code", the only changes are 3 new files added to the models directory

[glenn-jocher]

@jere357 thank you for your contribution to the YOLOv5 repository. We appreciate your effort in providing lighter config files. Our team will review your PR and provide feedback as soon as possible. Keep up the great work!

[github-actions]

👋 Hello there! We wanted to let you know that we've decided to close this pull request due to inactivity. We appreciate the effort you put into contributing to our project, but unfortunately, not all contributions are suitable or aligned with our product roadmap.

We hope you understand our decision, and please don't let it discourage you from contributing to open source projects in the future. We value all of our community members and their contributions, and we encourage you to keep exploring new projects and ways to get involved.

For additional resources and information, please see the links below:

• Docs: https://docs.ultralytics.com
• HUB: https://hub.ultralytics.com
• Community: https://community.ultralytics.com

Thank you for your contributions to YOLO 🚀 and Vision AI ⭐

[github-actions]

All Contributors have signed the CLA. ✅
_{Posted by the CLA Assistant Lite bot.}

[jere357]

I have read the CLA Document and I sign the CLA

[jere357]

@glenn-jocher wanna do anything about these lighter configs?
I talked to Ayush on twitter and he said he'd be ok with including this in the codebase but that i should run some benchmarks so it can be included in the docs, what do you think about that?