license: mit
Food Vision with EfficientNet
This repository contains the code for the Food Vision project using EfficientNet. The project involves building a deep learning model to classify food images into 101 different classes using the Food101 dataset.
Project Overview
This project utilizes TensorFlow and EfficientNet for image classification. It involves training a model on the Food101 dataset, fine-tuning the model, and evaluating its performance.
Dataset
The Food101 dataset is used for this project. It consists of 101,000 images across 101 food classes.
Data Preprocessing
The dataset is preprocessed using TensorFlow Datasets (TFDS). Images are resized, normalized, and batched to create an efficient input pipeline for the model.
Model Architecture
The EfficientNetV2B0 architecture is used as the base model for feature extraction. The top layers are added for classification. The model is compiled with a suitable loss function, optimizer, and metrics.
Training
The model is trained on the preprocessed data, and the training process is logged using TensorBoard. Checkpoints are saved to monitor the model's progress.
Fine-tuning
After feature extraction, the model is fine-tuned on the entire Food101 dataset. Learning rate reduction and early stopping callbacks are used to optimize training.
Results
The model's performance is evaluated on the test set, and the results are compared before and after fine-tuning.