|
--- |
|
license: apache-2.0 |
|
base_model: google/vit-base-patch16-224-in21k |
|
tags: |
|
- generated_from_trainer |
|
datasets: |
|
- image_folder |
|
metrics: |
|
- accuracy |
|
model-index: |
|
- name: Facial Expression Recognition |
|
results: |
|
- task: |
|
name: Image Classification |
|
type: image-classification |
|
dataset: |
|
name: image_folder |
|
type: image_folder |
|
config: default |
|
split: train |
|
args: default |
|
metrics: |
|
- name: Accuracy |
|
type: accuracy |
|
value: 0.8571428571428571 |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# Vision Transformer (ViT) for Facial Expression Recognition Model Card |
|
|
|
## Model Overview |
|
|
|
- **Model Name:** [motheecreator/vit-Facial-Expression-Recognition](https://huggingface.co/motheecreator/vit-Facial-Expression-Recognition) |
|
|
|
- **Task:** Facial Expression/Emotion Recognition |
|
|
|
- **Datasets:** [FER2013](https://www.kaggle.com/datasets/msambare/fer2013), [MMI Facial Expression Database](https://mmifacedb.eu) |
|
|
|
- **Model Architecture:** [Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit) |
|
|
|
- **Finetuned from model:** [vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) |
|
|
|
- Loss: 0.4353 |
|
- Accuracy: 0.8571 |
|
|
|
## Model description |
|
|
|
The vit-face-expression model is a Vision Transformer fine-tuned for the task of facial emotion recognition. |
|
|
|
It is trained on the FER2013 and MMI facial Expression datasets , which consists of facial images categorized into seven different emotions: |
|
- Angry |
|
- Disgust |
|
- Fear |
|
- Happy |
|
- Sad |
|
- Surprise |
|
- Neutral |
|
|
|
## Data Preprocessing |
|
|
|
The input images are preprocessed before being fed into the model. The preprocessing steps include: |
|
- **Resizing:** Images are resized to the specified input size. |
|
- **Normalization:** Pixel values are normalized to a specific range. |
|
- **Data Augmentation:** Random transformations such as rotations, flips, and zooms are applied to augment the training dataset. |
|
|
|
## Intended uses & limitations |
|
|
|
More information needed |
|
|
|
## Training and evaluation data |
|
|
|
More information needed |
|
|
|
## Training procedure |
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 5e-05 |
|
- train_batch_size: 32 |
|
- eval_batch_size: 32 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 4 |
|
- total_train_batch_size: 128 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_ratio: 0.1 |
|
- num_epochs: 10 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Accuracy | Validation Loss | |
|
|:-------------:|:-----:|:----:|:--------:|:---------------:| |
|
| 0.7964 | 1.0 | 798 | 0.7271 | 0.7869 | |
|
| 0.6567 | 2.0 | 1596 | 0.7380 | 0.7539 | |
|
| 0.6842 | 3.0 | 2394 | 0.7837 | 0.6287 | |
|
| 0.5242 | 4.0 | 3192 | 0.7839 | 0.6282 | |
|
| 0.4321 | 5.0 | 3990 | 0.7823 | 0.6423 | |
|
| 0.3129 | 6.0 | 4788 | 0.7838 | 0.6533 | |
|
| 0.4245 | 7.0 | 5586 | 0.8542 | 0.4382 | |
|
| 0.3806 | 8.0 | 6384 | 0.8531 | 0.4375 | |
|
| 0.3112 | 9.0 | 7182 | 0.8557 | 0.4372 | |
|
| 0.2692 | 10.0 | 7980 | 0.8571 | 0.4353 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.36.0 |
|
- Pytorch 2.0.0 |
|
- Datasets 2.1.0 |
|
- Tokenizers 0.15.0 |
|
|
