Hugging Face's logo

Spaces:
Adedoyinjames
/
Geodect
No application file

App Files Community
Geodect / README.md
Adedoyinjames's picture
Adedoyinjames
Update README.md
8e8b29d verified
preview code
|
raw
history blame contribute delete
2.2 kB

A newer version of the Gradio SDK is available: 6.11.0

Upgrade
metadata 
title: Geodect
emoji: 💻
colorFrom: gray
colorTo: green
sdk: gradio
sdk_version: 6.9.0
app_file: app.py
pinned: false
license: mit
short_description: AI-assisted computer vision for 3D geological modeling.

⚒️ Geodect: Geological 3D & Mapping

Geodect is an AI-assisted computer vision tool designed to convert field video of geological surfaces or rock samples into clean 3D models and high-resolution 2D orthomosaic maps.

🌟 Key Features

  • 2D Orthomosaic Mapping: Automatically extract, align, and stitch video frames into a single, high-resolution undistorted map.
  • 3D Sample Reconstruction: Transform video orbits of rock samples into interactive .obj meshes using Structure from Motion (SfM).
  • Intelligent Background Removal: Designed to segment and remove unwanted objects (like hands or tools) to focus on the geological surface.
  • CPU Optimized: Built to run efficiently within a Hugging Face Space without requiring high-end GPU resources.

🚀 Technical Workflow

  1. Extraction: OpenCV extracts optimal frames from uploaded field video.
  2. Processing: * Orthomosaic Mode: Detects features via SIFT/ORB and estimates homography for seamless stitching.
    • 3D Mode: Estimates camera poses and generates a dense point cloud converted into a surface mesh via Trimesh.
  3. Visualization: Interactive 3D rendering and high-res image downloads provided via the Gradio interface.

📂 Architecture

This project is implemented as a lightweight four-file Python application:

  • app.py: The main Gradio interface and pipeline controller.
  • Orthomosaic.py: The 2D stitching and blending engine.
  • ThreeDimagery.py: The 3D reconstruction and SfM module.
  • requirements.txt: Environment dependencies including OpenCV, SciPy, and Trimesh.

🤝 Credits & Acknowledgments

Developer: Adedoyin Ifeoluwa James

X: https://x.com/IAdedoyin64700

portfolio website: https://adedoyinjames-portfolio.vercel.app

blog: https://theadedoyinjournal.wordpress.com

Technical Resources: This project was developed with the technical support and computational resources of NORA RESEARCH LAB.