Available
FoodAI — Emotion-Based Food Intelligence
Initializing FoodAI Engine
Live AI System — FER-2013 Trained

Your Mood.
Your Food.
Powered by AI.

Real-time emotion detection meets smart food intelligence. Feel happy, eat something delicious.

▶ Try Emotion Scan ⭐ View on GitHub
0
Training Samples
0
Emotion Classes
~70%
Test Accuracy
∞
Meal Combos
FoodAI — Live Detection
😊
Happy Detected
😊 Happy
87%
😲 Surprise
8%
😐 Neutral
5%
Emotion Engine

How Do You Feel Right Now?

Select your current emotion and watch the AI recommend the perfect meal — live, dynamic, personalized.

AI Architecture

How the Neural Brain Works

A deep pipeline of CNN layers, reinforcement learning, and smart recommendation logic working in perfect sync.

🧠
CNN Emotion Model
3 convolutional + 2 dense layers trained on FER-2013 (25k samples). Classifies 7 emotions with ~70% accuracy.
🔄
Reinforcement Learning
Q-learning feedback loop. Every user rating refines the recommendation policy over time.
🍽️
Smart Recommender
Maps detected emotions to curated food profiles with confidence scoring and pairing logic.
👁️
Computer Vision
OpenCV handles real-time face detection and preprocessing before the CNN inference step.
Dataset & Training

Trained on 30,000+ Human Faces

FER-2013 dataset with 7 emotion categories, augmented for real-world robustness.

😊 Happy~7,000
😐 Neutral~6,000
😢 Sad~5,000
😠 Angry~4,000
😨 Fear~4,000
😲 Surprise~4,000
🤢 Disgust~400
01
Load FER-2013 Dataset
Scans emotion folders, loads 48×48 grayscale face images with labels.
02
Data Augmentation
Random rotation, horizontal flip, zoom — prevents overfitting.
03
Build CNN Architecture
3 Conv2D + MaxPool + Dropout + 2 Dense layers with softmax output.
04
Train with Callbacks
Early stopping + learning rate scheduling for optimal convergence.
05
Save Model
Best weights saved to model/emotion_cnn.h5 for inference.
Technology

Built With World-Class Tools

Production-grade AI/ML stack, from raw pixels to personalized plates.

🤖
AI / ML
TensorFlow / Keras
Deep learning framework powering the CNN emotion model with 3 convolutional layers and softmax classification.
👁️
Computer Vision
OpenCV
Real-time face detection, image preprocessing, and frame capture before neural network inference.
🧮
Numerical Computing
NumPy
Matrix operations, data normalization, and array manipulation throughout the ML pipeline.
🌐
Backend
Flask
Lightweight web server and REST API exposing emotion detection and food recommendation endpoints.
🔄
Reinforcement Learning
Custom Q-Learning
Self-improving recommendation engine that learns from user ratings via Q-table updates.
🐍
Runtime
Python 3.8+
Core runtime and package ecosystem. Everything from training to serving runs in Python.
Real-World Impact

Where FoodAI Comes Alive

🍕
Food Tech
Smart Restaurants
Emotion-aware digital menus that adapt to how guests feel the moment they walk in.
🏥
Healthcare
Mood Nutrition
Clinical dietary plans tailored to emotional states for better mental and physical health outcomes.
🎓
Education
ML Teaching Tool
An interactive showcase of CNN, computer vision, and reinforcement learning in one live system.
📱
Mobile Apps
Emotion Food Diary
Track how your emotional states influence eating patterns over time with AI-powered insights.
🎯
Marketing
Emotional Advertising
Serve food ads matched to viewer emotions for radically higher engagement and conversion.
🏋️
Wellness
Mood-Based Nutrition
Wellness platforms that prescribe foods for emotional regulation and cognitive performance.
🍕
Food Tech
Smart Restaurants
Emotion-aware digital menus that adapt to how guests feel the moment they walk in.
🏥
Healthcare
Mood Nutrition
Clinical dietary plans tailored to emotional states for better mental and physical health outcomes.
🎓
Education
ML Teaching Tool
An interactive showcase of CNN, computer vision, and reinforcement learning in one live system.
📱
Mobile Apps
Emotion Food Diary
Track how your emotional states influence eating patterns over time with AI-powered insights.
🎯
Marketing
Emotional Advertising
Serve food ads matched to viewer emotions for radically higher engagement and conversion.
🏋️
Wellness
Mood-Based Nutrition
Wellness platforms that prescribe foods for emotional regulation and cognitive performance.
Project Architecture

Clean, Modular Codebase

Click folders to explore the project structure interactively.

ID
Isfaq Evan Dipro
AI Specialist · Full-Stack Developer

Founder of Brainstation Cyber Hub (BSCH). Building intelligent, production-ready systems where AI meets real-world experience design.

🌐 Portfolio ⭐ GitHub Repo 📧 Contact
TensorFlow Computer Vision Flask Python Full-Stack BSCH