This repository contains a project focused on classifying images of 10 different food classes using deep learning and computer vision techniques. The project leverages transfer learning and fine-tuning with TensorFlow to achieve a high accuracy of 90% on the full test set. The implementation is carried out on Google Colab for efficient computation and easy accessibility.
- Data Source: Preprocessed food image datasets derived from the 'Organizing_The_Food101_Dataset' notebook.
- Deep Learning Framework: TensorFlow
- Model Type: Deep Learning, Computer Vision, Transfer Learning Feature Extraction, and Transfer Learning Fine-Tuning with Data Augmentation
- Goal: Classifying images into 10 different food classes
- Accuracy: Achieves 90% accuracy on the full test set
- Environment: Google Colab for development and execution
- Notebooks:
- Organizing_The_Food101_Dataset: Notebook for preparing and organizing the Food-101 dataset from Kaggle.
- Food Image Deca (10 Classes) Classifier - Computer Vision: Notebook detailing the data loading and preprocessing, model training, evaluation, and results.
- Scripts:
- helper_functions.py: Contains utility functions needed for the project, located in the 'scripts' folder.
- Data: Preprocessed datasets available through publicly accessible Google Drive links in the 'Food Image Deca (10 Classes) Classifier - Computer Vision' notebook.
- Clone the repository:
git clone! https://github.com/IsraelAzoulay/food-image-deca-classifier-computer-vision.git
- Open the provided Google Colab notebooks: Navigate to the 'notebooks' folder and open the desired notebook in Google Colab.
- Run the Notebooks: Follow the instructions in the notebooks to download the datasets, preprocess the data, train and evaluate the models.
- Customize: Feel free to modify the models and code for your experiments.
Contributions are welcome! Please feel free to submit a Pull Request.
This project is licensed under the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) License.