It is a realtime expression detector using MediaPipe, Python & Machine learning algorithms.
The realtime expression detector is designed using Media Pipe and Python. Users can be able to leverage their webcam to decode what their body language says at a point in time. So, specifically in order to do this, the detector is leveraging pre-trained data as well as a custom machine learning model to be able to take the landmarks from user's face as well as different poses from their body.
OpenCV and CSV are used to capture the landmarks related to classified facial and gesturing expressions. The collected data are stored in the spreadsheet which are used to train the machine to assume the real time detected poses. The more we collect landmarks for different body and face gestures, the more we get the accurate prediction results.
I have used the following 4 machine learning models for expression detection.
- Logistic Regression
- Ridge Classifier
- Random Forest
- Gradient Boosting
- Language: Python
- Platform: Jupyter Notebook IDE
- Packages: MediaPipe, cv2, csv, os, numpy, pandas, sklearn, train_test_split, pickle