CLIP_Emotions

CLIP-ViT-Large-Patch14 Fine-Tuned on FER-2013 Dataset

---

license: apache-2.0

This model is a fine-tuned version of the CLIP-ViT-Large-Patch14 model on the FER-2013 dataset.

Overview

The model is designed for facial emotion recognition and can classify images into the following 7 primary emotions:

• Neutral
• Anger
• Happiness
• Fear
• Disgust
• Sadness
• Surprise

It leverages the powerful vision encoder from the original CLIP model, developed by OpenAI to enable zero-shot image classification. The model has been further fine-tuned specifically for facial emotion recognition using the FER-2013 dataset.

Important Note: This model should be used strictly for research purposes. Deployment in real-world or commercial applications should involve thorough domain-specific testing and fairness evaluation due to potential biases and performance variability.

Model Details

• Base Model: openai/clip-vit-large-patch14
• Fine-tuned on: FER-2013
• Task: Facial Emotion Recognition
• Number of Classes: 7

Intended Use

This model is intended for researchers and developers working on:

• Facial expression recognition
• Emotion detection in images
• Human-computer interaction studies
• Psychological and behavioral modeling

Limitations

• The model was trained exclusively on static grayscale face images aligned in frontal pose.
• Performance may degrade significantly with occluded faces, side profiles, or low-resolution images.
• The model has not been evaluated for fairness across different demographics such as race, gender, or age groups.
• It may exhibit bias depending on how class labels are defined and constructed.