ravi86/mood_detector

🎭 Face Expression Detector

A deep learning model that classifies facial expressions in grayscale images into one of seven core emotions. Designed for applications in emotion analytics, human-computer interaction, and psychological research.

---

📊 Model Overview

This model takes 48x48 grayscale face images and classifies them into:

• 😠 Angry
• 🤢 Disgust
• 😨 Fear
• 😄 Happy
• 😢 Sad
• 😲 Surprise
• 😐 Neutral

Dataset: FER2013
Training Samples: 28,709
Testing Samples: 3,589

---

🧠 Model Architecture

• 📦 Custom CNN
  • 3 Convolutional Layers
  • Batch Normalization
  • ReLU Activation
  • Dropout for regularization
• 📈 Optimizer: Adam
• 🔥 Loss Function: Categorical Crossentropy
• ⏱️ Epochs: 100

---

✅ Performance

📌 Add your actual performance metrics here:

• Accuracy on FER2013 Test Set: ~1.0%

---

🗂️ Required Files

• model.h5 → Model Weights
• config.json → Configuration file (Transformers-based)
• preprocessor_config.json → Preprocessing setup (if needed)
• requirements.txt → Python dependencies

---

🚀 Use Cases

• 🎮 Real-time emotion feedback in games or virtual assistants
• 🎓 Emotion analysis for psychological and behavioral studies
• 🎥 Enhancing video-based UX with dynamic emotion tracking

---

⚠️ Limitations

• Works best with centered 48x48 grayscale faces
• Face detection (e.g., MTCNN) required before prediction
• FER2013's demographic diversity is limited → potential bias

---

⚙️ Installation

Follow these steps to set up the environment and dependencies: --pip install -r requirements.txt torch>=1.9.0 transformers>=4.20.0 pillow>=8.0.0

1. Clone the Repository

git clone https://github.com/TRavi8688/Mood-Based-Music-Player cd mood_detector

##🧪 How to Use (Transformers-based)

Follow these steps to preprocess an image and predict facial expression using the pre-trained Transformers-based model: Python

from transformers import AutoModelForImageClassification, AutoImageProcessor
from PIL import Image
import torch

1. Load Model and Preprocessor

# STEP 1: Install dependencies
!pip install tensorflow pillow numpy

# STEP 2: Download model file using `requests`
import requests

model_url = "https://huggingface.co/ravi86/mood_detector/resolve/main/my_model.h5"
model_path = "my_model.h5"

# Download the file
response = requests.get(model_url)
with open(model_path, "wb") as f:
    f.write(response.content)

print("✅ Model downloaded successfully!")

2. Load and Preprocess the Image

image_path = "your_image.jpg"  # 🔁 Replace with your image file
image = Image.open(image_path).convert("L").resize((48, 48))  # Convert to grayscale and resize

3. Make Predictions

outputs = model(**inputs)
probs = torch.softmax(outputs.logits, dim=-1)  # Convert logits to probabilities
predicted_class = probs.argmax().item()  # Get the predicted class index

4. Interpret the Result

emotions = ["Angry", "Disgust", "Fear", "Happy", "Sad", "Surprise", "Neutral"]
print(f"Predicted Emotion: {emotions[predicted_class]}")

☁️ Deploy to Hugging Face Hub

Use these commands to prepare and push your model to the Hugging Face Hub:
Bash

# Step 1: Install & Login
pip install huggingface_hub
huggingface-cli login

from huggingface_hub import upload_folder

upload_folder(
    folder_path="path/to/mood_detector",
    repo_id="ravi86/mood_detector",
    repo_type="model",
    commit_message="🚀 Upload mood detection model"
)

🧭 Ethical Considerations

⚖️ Bias: The FER2013 dataset may exhibit biases in demographic representation. Exercise caution when interpreting results across diverse populations.
🔒 Privacy: Ensure strict compliance with data privacy laws (e.g., GDPR, CCPA) when using this model on personal or sensitive images. Do not use without explicit consent.
❗ Misuse: This model is not intended for unauthorized surveillance, profiling, or any other unethical applications.

👤 Contact 📬 For questions, support, or collaborations: Hugging Face → @ravi86 Gmail→ [email protected] ⭐ If you find this project useful, consider giving a star or contributing!