Food Vision 🍔 📷

FoodVision is a CNN Image Classification Model that classifies images into 101 food categories using the Food101 dataset. It can identify 101 different food classes.

It is based upon a pre-trained Model that has been fine-tuned on the Food101 Dataset.

Fun Fact

The Model actually beats the DeepFood Paper's model which also trained on the same dataset.

The Accuracy aquired by DeepFood was 77.4% and our model's 85% . Difference of 8% ain't much, but the interesting thing is, DeepFood's model took 2-3 days to train while our's barely took 90min.

Dataset used : Food101

Model Used : EfficientNetB1

Accuracy : 85%

Check the deployed app

app_video.mp4

Once an app is loaded,

1. Upload an image of food.
2. Once the image is processed, Predict button appears. Click it.
3. Once you click the Predict button, the model prediction takes place and the output will be displayed along with the model's Top-5 Predictions
4. And voilà, there you go.

Model Training

If you want to know how the model was trained check out FoodVisionGithub.ipynb Notebook

Breaking down the repo

At first glance the files in the repo may look intimidating and overwhelming. To avoid that, here is a quick guide :

• .gitignore : tells what files/folders to ignore when committing
• app.py : Our Food Vision app built using Streamlit
• utils.py : Some of used fuctions in app.py
• foodvisiongithub.ipynb : Notebook used to train the model
• model/ : Contains all the models used as .hfd5 files
• requirements.txt : List of required dependencies required to run app.py
• extras/ : contains images and files used to write this README File

Inspired by Daniel Bourke's Course