This project is a real-time emotion detection system that uses your webcam feed to detect human faces and analyze their emotions. It leverages OpenCV for face detection and DeepFace for emotion recognition powered by deep learning.
This is a lightweight prototype of how computer vision and deep learning can be integrated to build emotion-aware systems such as smart surveillance, human-computer interaction tools, or sentiment-aware applications.
- Video Feed Capture: The webcam is accessed using OpenCV's
VideoCapture. - Face Detection: Faces are identified using Haar Cascade classifiers.
- Emotion Analysis: Detected faces are passed to the DeepFace library which uses a pre-trained deep neural network to classify emotions (e.g., happy, sad, angry, etc.).
- Display Output: Bounding boxes and the predicted dominant emotion are overlaid on the video stream in real time.
⚠️ Python 3.7 – 3.10 is recommended due to DeepFace's dependencies.
git clone https://github.com/yourusername/emotion-detector
cd emotion-detectorpython -m venv venv
source venv/bin/activate # or venv\Scripts\activate on Windowspip install -r requirements.txtpython main.py- The webcam window will pop up.
- Real-time bounding boxes and emotions will be displayed on detected faces.
- Press
qto exit the program.
This project taught me the following:
-
✅ How to use OpenCV for real-time video capture and face detection.
-
✅ How to integrate DeepFace for high-level emotion recognition using deep learning models.
-
✅ How to handle frame-by-frame analysis efficiently and deal with practical runtime issues like:
- Missing Haar cascade files
- TensorFlow GPU warnings (I will recomend u install tensorFlow -CPU version for not getting Cuda errors)
- Frame synchronization
-
✅ Debugging common computer vision issues (e.g., wrong image formats, empty classifier errors).
-
✅ Importance of modularizing code for readability and robustness.
- Use a more efficient face detector like MTCNN or RetinaFace
- Store emotion logs for analysis over time