|
|
--- |
|
|
base_model: microsoft/resnet-50 |
|
|
tags: |
|
|
- image-classification |
|
|
- ferplus |
|
|
- emotions |
|
|
--- |
|
|
|
|
|
# Fine-Tuned ResNet-50 on FERPlus Dataset |
|
|
|
|
|
This model is a fine-tuned version of ResNet-50 on the [FERPlus dataset](https://www.kaggle.com/datasets/arnabkumarroy02/ferplus), which is more balanced ferplus dataset as owner claimed. |
|
|
|
|
|
## Model Details |
|
|
- **Base Model**: [Microsoft ResNet-50](https://huggingface.co/microsoft/resnet-50) |
|
|
- **Dataset**: FERPlus, which contains grayscale images of faces labeled with emotion categories. |
|
|
- **Task**: Emotion Classification |
|
|
- **Labels**: |
|
|
- 0: Angry |
|
|
- 1: Contempt |
|
|
- 2: Disgust |
|
|
- 3: Fear |
|
|
- 4: Happy |
|
|
- 5: Neutral |
|
|
- 6: Sad |
|
|
- 7: Surprise |
|
|
|
|
|
## Preprocessing Details |
|
|
This model was fine-tuned on FERPlus dataset images resized to 224x224 pixels. Standard data augmentation techniques were applied, and normalization was performed with the following values: |
|
|
- **Mean**: `[0.485, 0.456, 0.406]` |
|
|
- **Standard Deviation**: `[0.229, 0.224, 0.225]` |
|
|
|
|
|
### Training Hyperparameters |
|
|
- **Batch Size**: 16 |
|
|
- **Epochs**: 10 |
|
|
- **Learning Rate**: 2e-5 |
|
|
- **Weight Decay**: 0.01 |
