FoodSnap.ai

FoodSnap.ai is a complete AI-powered food analysis platform that delivers a seamless pipeline — from data collection and intelligent inference to real-time visualization. It combines food image classification, weight estimation, and nutritional analysis into a unified experience with a rich, user-friendly interface.

1. Food Classification Model — Classifies Indian food images into predefined categories.
2. Weight Prediction Model — Predicts the total weight (mass in grams) of food items from images.
3. Inference Server — REST API for real-time predictions using both models.
4. FoodSnap Web App — Provides a user-friendly interface and seamless experience.
5. Data Scrapper — Python script that executes API calls to extract nutritional data and store it as CSV.

🎬 Demo Preview

Demo Link

---

1. Food Classification Model

Dataset

Indian Food Classification (Kaggle)

Overview

• Uses MobileNetV2 as a base model with transfer learning.
• Can classify below food items

["burger", "butter_naan", "chai", "chapati", "chole_bhature", "dal_makhani", "dhokla", "fried_rice", "idli", "jalebi", "kaathi_rolls", "kadai_paneer", "kulfi", "masala_dosa", "momos", "paani_puri", "pakode", "pav_bhaji", "pizza", "samosa"].

Evaluation

• Test accuracy : 87.39%

---

2. Food Weight Prediction Model

Dataset

Food Nutrients Dataset (HuggingFace)

Overview

• Model is a CNN-based regression network.
• Predicts food weight (total mass in grams) from images.

Evaluation

• RMSE (Root Mean Squared Error) : 131.78434
• R² Score : 0.29374
• Null/Baseline Model Test RMSE : 156.81286

---

3. Inference Server (Flask)

A simple REST API server that takes an image and returns:

• Predicted food name
• Class code
• Estimated weight (grams)
• Confidence score (%)
• Time taken (seconds)

Endpoint

POST /predict

Request

• multipart/form-data
• Key: foodImage
• Value: food image file (.jpg, .png, etc.)

Response

{
  "food_name": "Masala Dosa",
  "food_code": "masala_dosa",
  "weight": 322.45,
  "confidence": 94.38,
  "time_taken": "0.86 seconds"
}

Setup & Installation

If you encounter any version-related issues with Python & tensorflow, switch to Python 3.11

1. Clone the repository

git clone <repo url>
cd Food_Classification_Model

2. Install dependencies

pip install -r req.txt

3. Run script

python3.11 server/app.py

---

4. FoodSnap Web App — Laravel + FilamentPHP

A powerful and modern web application built using Laravel and FilamentPHP to manage, visualize, and interact with AI-powered food classification and weight estimation models.

This panel acts as the front-end interface for:

• Uploading food images for inference*
• Displaying classification results and estimated weight
• Showing food nutrition details
• Presenting charts and statistical overviews
• Managing user-submitted entries and food records

* Only food items with a confidence level of 60% or higher are displayed.

---

Requirements

• PHP >= 8.2
• Composer
• MySQL or SQLite
• Node.js & npm

---

Installation Guide

1. Clone the Repository

git clone <repo url>
cd FoodSnap_App

2. Install PHP Dependencies

composer install

---

3. Create and Configure .env

cp .env.example .env
php artisan key:generate

Update your .env file with the correct DB and app settings:

DB_CONNECTION=mysql         # or sqlite
DB_HOST=127.0.0.1
DB_PORT=3306
DB_DATABASE=your_database
DB_USERNAME=your_username
DB_PASSWORD=your_password


FOOD_DETECTION_API_URL='http://localhost:5000/predict'

---

4. Run Migrations

php artisan migrate

---

5. Seed Initial Data

php artisan db:seed

---

6. Create Storage Link

php artisan storage:link

7. Serve the App

php artisan serve

Your app will now be accessible at: http://localhost:8000

---

5. Nutrition Info Scraper

This Python script mimics an API request made by a food website to fetch nutritional data for a list of specified foods. It extracts structured nutritional details and saves them to a CSV file for further analysis or integration.

What It Does

• Performs API call to a food website.
• Sends food names via POST request to the nutrition endpoint.
• Parses the response to extract:
  • Food Name
  • Serving Size
  • Calories
  • Protein
  • Carbohydrates
  • Fats
  • Sugar
  • Fibre
• Saves all data to a .csv file.

Installation

1. Clone the repository

git clone <repo url>
cd Data_Scrapper

2. Install dependencies

pip install -r req.txt

3. Run script

python scrapper.py