Micropix-Twissie@Twistcode

This is a submission repository for European Association Of Geoscientist & Engineers (EAGE) Earth-Hack Hackathon 2019

Description

Rock properties detection from core images

We try to analyze and extract rock properties by classifying the rock properties and location corresponding to the core images using transfer learning.

Members

• Akmal
• Hasif
• Sidiq
• Syuhada

Built With

Machine Learning Library

• Pytorch
• FastAI
• Scikit-learn

Application Setup

• Install all the library required

pip3 install flask torch torchvision fastai scikit-learn numpy scipy pandas

• Download the trained models and data from google drive
  • https://1drv.ms/f/s!ArJJyxURU-h_h5QECCEsR1SG6cGZdQ files to src/satellite/models
  • https://1drv.ms/f/s!ArJJyxURU-h_h5QDNLfWbj2uhY54aw files to src/satellite/train
• Run the script to initialize the server

python3 server.py

• To access the application http://localhost:5000
• To access the API, send an image named 'file' as a multipart/form-data POST request to http://localhost:5000/image

Demo

Application Demo Link

License

The hack is governed by Creative Commons, CC BY. You can:

• Copy
• Adapt or Modify
• Redistribute (Publish, Display, Publicly Perform or Communicate the Work
• License to others