Hugging Face's logo

Spaces:
rmayormartins
/
yolov5-sunspot-hunter
Sleeping

App Files Community
yolov5-sunspot-hunter / README.md
rmayormartins's picture
rmayormartins
Subindo arquivos
8d9557a
preview code
|
raw
history blame contribute delete
2.7 kB

A newer version of the Gradio SDK is available: 5.44.0

Upgrade
metadata
title: Yolov5 Sunspot Hunter
emoji: 🌟
colorFrom: green
colorTo: blue
sdk: gradio
sdk_version: 4.12.0
app_file: app.py
pinned: false
license: ecl-2.0

YOLOv5 Sunspot Hunter 🌟

Explore the dynamic solar surface with the YOLOv5 Sunspot Hunter! This application is designed to detect and analyze sunspots using state-of-the-art object detection technology.

Developer

Developed by Ramon Mayor Martins (2023)

About the Project

This tool using YOLOv5, an advanced neural network, for the detection and classification (hunting) of sunspots. The sunspot images were collected from several high-quality sources, including the SOHO satellite, and other NASA and ESA archives, under Creative Commons licenses. These images were then annotated with precision using Makesense.ai.

Key Features

  • Image Source: The sunspot images were sourced from SOHO satellite, NASA, and ESA archives.
  • Labeling: Annotations were done using Makesense.ai.
  • Model Training: The model was trained with YOLOv5, achieving satisfactory metrics including mAP (mean Average Precision).
  • Model File: The 'best.pt' file is used, which represents the model's optimal state after training.

How to Use

  • Launch: Start the YOLOv5 Sunspot Hunter by running the app.py script in Gradio.
  • Image Upload: Users can upload their own images of the sun or utilize current solar images from websites like Space Weather Live, LMSAL, SOHO Realtime Images, and The Sun Today.
  • Detection: The tool processes the uploaded image, identifying and highlighting sunspots with high precision.
  • Results: View the results instantly, with sunspots clearly marked and classified.

Feedback and Contributions

Feel free to reach out or contribute to the project. Your feedback and contributions are highly appreciated!

License

This project is released under the ECL-2.0 license.

Acknowledgments

Special thanks to the teams at NASA, ESA, and SOHO for providing valuable solar data.

Check out the configuration reference at Hugging Face Spaces Config Reference.