FLEC - Facial Landmark Emotion Classification Project
Welcome to the Facial Landmark Emotion Classification (FLEC) project! Our goal is to develop a robust classifier capable of recognizing six facial expressions using the 68 facial landmarks provided in .csv files.
Project Overview :mag:
Step 1: Characterizing Expressions 🎯
In this crucial step, we aim to predict the emotion column in the emotion.csv file based on the facial points available in the SXXX/omlands.csv files. Our approaches include:
- Utilizing the coordinates of facial points.
- Analyzing the movement of points between neutral and apex emotion images.
To enhance accuracy, we’ll experiment with face alignment techniques, exploring both the use of raw points and a common frame of reference.
Step 2: Handling Imbalance 🙌
Given the highly imbalanced dataset, our second phase focuses on creating a balanced dataset. We’ll assess the impact on results compared to the initial configuration, ensuring more reliable and unbiased model training.
Step 3: Study of Occlusions and Noises 🕶️
Understanding the impact of occlusions and noises on facial landmarks is crucial for real-world applications.
Step 3.1: Creating Occlusions and Noises 👥
We’ll simulate various occlusions and noises, starting from small regions (e.g., eyes, eyebrows) to larger occlusions. This step aims to evaluate model robustness under different alteration scenarios.
Step 3.2: Evaluating Robustness 📊
Our evaluation will provide insights into how well our learning techniques handle occlusions and noises. Quantification and appropriate measurements will guide our assessment of robustness.
Project Report 🛠️
Following is the official report for the project.
Contributing 🤝
Contributions are highly encouraged! If you have suggestions, improvements, or feature requests, feel free to open an issue or create a pull request on our public repository.
License 📝
This project is licensed under the MIT License - see the LICENSE file for details.
Developed by Pierre LAGUE and François MULLER (@franzele21) at the University of Lille, France. 🚀📊