dima806's picture
Update README.md
f7836c6 verified
|
raw
history blame
1.2 kB
metadata
license: apache-2.0
metrics:
  - accuracy
  - f1
base_model:
  - google/vit-base-patch16-224-in21k
pipeline_tag: image-classification
library_name: transformers
datasets:
  - nateraw/fairface

Detects age group with about 58% accuracy based on an image.

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

image/png

Classification report:

              precision    recall  f1-score   support

         0-2     0.8065    0.6944    0.7463       180
         3-9     0.7713    0.8207    0.7952      1249
       10-19     0.5117    0.3619    0.4239      1086
       20-29     0.6241    0.7356    0.6753      3026
       30-39     0.4877    0.4917    0.4897      2099
       40-49     0.4860    0.4063    0.4426      1238
       50-59     0.5035    0.5021    0.5028       725
       60-69     0.4593    0.4930    0.4755       286
more than 70     0.0000    0.0000    0.0000       111

    accuracy                         0.5809     10000
   macro avg     0.5167    0.5006    0.5057     10000
weighted avg     0.5674    0.5809    0.5708     10000