The objective is to determine a person's head orientation in images or videos by predicting the three-dimensional rotation angles of Pitch, Yaw, and Roll.
This project focuses on developing a head pose estimation system using MediaPipe and the AFLW2000 dataset. By leveraging MediaPipe's pre-trained models and the rich annotations of 3D facial landmarks in the AFLW2000 dataset, the goal is to accurately predict the orientation of human heads in images or videos. Key steps include data preprocessing to clean and prepare the dataset, model development by training and fine-tuning machine learning models within the MediaPipe framework using GridSearchCV, and integration for real-world applications.