This machine learning web-application was built based on a small dataset (1000 pictures, with 5 categories: amethyst, ammonite, aventurin, empty, obsidian), and its purpose is to recognize minerals and fossils from a personal collection.
Open a new terminal:
1.Go to your Downloads location (cd Downloads)
2.Clone this repository (git_clone [email protected]:IonelS-coder/predict_geological_sample.git)
3.Open the predict_geological_sample folder (cd predict_geological_sample)
4.Install the necessary libraries (ideally in a newly created conda environment:
4.1 Create and activate the conda environment
conda create -n predict_geological_sample pip python=3.8
conda activate predict_geological_sample
pip install -r requirements.txt
5.Check that all the necessary libraries were installed (pip list)
6.Access streamlit folder (cd streamlit) and run this command:
- streamlit run interactive_app_rocks.py
- open the link (e.g. http://localhost:8501/) in a browser.
- load one of the test images and check the results
Test and have fun with the app.
** Extra
For more details check the following Jupyter Notebook file* (*jupyter package is required): M3_TEST_Rock_mineral_image_classification_transfer_learning.ipynb
Check also the M3_presentation.slides.html file from the reveals.js folder for a presentation of the project