🧠 AffectSense

AffectSense is a Convolutional Neural Network (CNN)-based model designed for emotion and affect recognition from visual or image-based data. The model leverages a pre-trained ResNet-50 backbone and has been fine-tuned for affective computing tasks such as emotion classification and mood detection.

🚀 Usage

You can load a model like this:

import torch
from torchvision import models

# Load the model (example if using torch.load)
model = torch.load("path_to_checkpoint.pth")
model.eval()

Or, if packaged in a model class:

from affectsense import AffectSenseModel

model = AffectSenseModel.from_pretrained("tawheed-tariq/AffectSense")

📊 Intended Uses & Limitations

Use Cases

• Emotion recognition from facial images
• Affective content tagging in videos
• Visual mood estimation
• Human-computer interaction systems

Limitations

• May not generalize well across unseen demographics or lighting conditions
• Not suitable for clinical diagnosis
• Accuracy depends on the diversity of training data

🏗️ Model Architecture

• Backbone: ResNet-50 (pre-trained on ImageNet)
• Modified Head: Custom classification head for emotion categories
• Input Size: Typically 224×224 RGB images

📁 Training Data

The models were trained on custom-curated datasets with emotion-labeled visual data. Examples include facial emotion datasets or affective scene datasets.

📜 License

This model is licensed under the Apache 2.0 License.

✍️ Citation

If you use this model in your research, please cite:

@misc{affectsense2025,
  title={AffectSense: CNN-based Emotion Recognition Model using ResNet-50},
  author={Tariq, Tavaheed},
  year={2025},
  howpublished={\url{https://huggingface.co/tawheed-tariq/AffectSense}},
}

---