---

Powered by Samueli924

Important

By using this repository or any code related to it, you agree to the MIT License. This is the author's only account and repository. To prevent impersonation or irresponsible actions, please comply with the MIT license this Repository uses.

Tip

The models contained are trained based on a train dataset of 80k 200x200 images. For now, it supports predicting gemstone shape, color and identification. Before clone the code and start using, you need to download the pretrained models from models.7z.

---

AutoGem is based on the latest Convolutional Neural Network. The train dataset sample and detailed training model will be made public soon. Please ⭐ star this repo to follow up.

📚 Getting Started quickly

1. Install requirements:
   Begin by pip/pip3 install -r requirements

2. Set up Tensorflow GPU(Optional)
   Follow the Official Documentation to configure GPU Support

3. Start FLASK App
   Start the app by python3 app.py

4. Use the AutoGemUI
   Use browser http://127.0.0.1:5000

5. Upload Colored Gemstone Image
   The gemstone must be in the .shape file within the downloaded models.7z file. The image uploaded should be 1:1 ratio, best to be 200x200 pixels.

💡 Examples of Prediction result