Analysis of Elicited and Acted Emotional Expressions in PEDFE

Group Members

• Zikun Fu
• Tony Wang

Objective

1. Investigate the differences between genuine and posed emotional expressions within the Padova Emotional Dataset of Facial Expressions (PEDFE).
2. Provide insights into the nuances of emotional expression and the ability of current technological systems to recognize genuine emotional states.

Research Question

1. How do automated emotion recognition systems perform in differentiating between genuine and posed emotional expressions in the PEDFE dataset?
2. How do genuineness, intensity, and hit rate correlate with the emotion classification accuracy, and are these correlations consistent between genuine and posed emotional expressions?

Methodology

1. Data Preparation: Utilize the PEDFE dataset for the analysis.
2. Feature Extraction: Use Py-Feat for extracting relevant features from the dataset.
3. Emotion Classification: Apply classification models to identify patterns and distinctions in facial emotions.
4. Result Analysis: Evaluate the classifier's performance and analyze the distinctions between genuine and posed expressions.

File Description

• Pyfeat.ipynb: Notebook used for feature extraction and emotion classification with Py-Feat.
• Analysis.ipynb: Notebook used for data processing and result analysis.
• /data:
  • PEDFE_set_clips: Contains modified video clips from the Padova Emotional Dataset of Facial Expressions (PEDFE), used for emotion analysis.
  • Supplemental_Material_T1.csv: Provides labels for the clips including information about the genuineness or type of emotion expressed.
  • combined_results.csv: Aggregated classification results, including mean scores of detected emotions for each video clips.

Installation

To set up the project environment and run the notebooks, follow these steps:

git clone https://github.com/ZikunFu/CSCI5730_GroupProj.git
cd CSCI5730_GroupProj
pip install -r requirements.txt

Credits

1. Miolla, Alessio, Matteo Cardaioli, and Cristina Scarpazza. "Padova Emotional Dataset of Facial Expressions (PEDFE): A unique dataset of genuine and posed emotional facial expressions." Behavior Research Methods 55.5 (2023): 2559-2574.
2. Muhammod, Rafsanjani, et al. "PyFeat: a Python-based effective feature generation tool for DNA, RNA and protein sequences." Bioinformatics 35.19 (2019): 3831-3833.
3. Jolly, E., Cheong, J. H., Xie, T., & Chang, L. J. (2022). Included pre-trained detectors. Py-Feat. Retrieved from https://py-feat.org/pages/models.html