README.md

---
license: apache-2.0
metrics:
- accuracy
- f1
base_model:
- google/vit-base-patch16-224-in21k
---
Returns age group with about 91% accuracy based on facial image.

See https://www.kaggle.com/code/dima806/age-group-image-detection-vit for more details.

![image/png](https://cdn-uploads.huggingface.co/production/uploads/6449300e3adf50d864095b90/Fp88lO_Z8KNt1JNzHyg1s.png)

```
Classification report:

              precision    recall  f1-score   support

      MIDDLE     0.8316    0.9278    0.8771      4321
       YOUNG     0.9598    0.8563    0.9051      4322
         OLD     0.9552    0.9477    0.9515      4322

    accuracy                         0.9106     12965
   macro avg     0.9155    0.9106    0.9112     12965
weighted avg     0.9155    0.9106    0.9112     12965
```