CSE-666 Programming Assignment 01
Detailed exploration of facial detection and analysis in a comprehensive programming assignment.
Course Details
- Course Number: CSE 666
- Assignment: Programming Assignment-01
- Professor: Nalini Ratha
Team Information
- Student: Ronak Haresh Chhatbar
- UBName: ronakhar
- Teammates: Ronak Haresh Chhatbar
Tasks and Techniques in CSE666 Assignment
Overview of the tasks and the computational techniques applied in the programming assignment.
Task 1: Annotation
Utilized LabelMe to annotate 129 images with bounding boxes for facial detection.
Task 2: Face Detection
Employed the MTCNN model from the face-image-analysis repository to detect faces in the 'congress.jpg' image, successfully identifying 133 faces.
Task 3: Sentiment/Expression Analysis
Applied the face-emotion-recognition library with mobile_net7 model to analyze and classify facial expressions into categories such as anger, disgust, happiness, etc.
Task 4: Gender Classification
Implemented gender classification using the simple_CNN.81-0.96.hdf5 TensorFlow model, achieving notable precision and recall metrics.
Task 5: Face Pose Estimation
Conducted face pose estimation using the Rotation Representation for Unconstrained Head Pose Estimation library to determine the orientation of faces.
Task 6: Feature Extraction
Extracted facial features using the ARC-face model, producing embeddings for facial recognition.
Task 7: Face Recognition
Matched faces against a dataset of lawmakers using embeddings, labeling recognized individuals or marking as "Unknown".
Technology and Tools
- LabelMe for Annotation
- MTCNN for Face Detection
- Face Emotion Recognition Library
- TensorFlow for Gender Classification
- Head Pose Estimation Library
- ARC-face Model for Feature Extraction