---
license: apache-2.0
metrics:
- accuracy
- f1
pipeline_tag: image-classification
base_model:
- google/vit-base-patch16-224-in21k
---
Detects garbage type with about 93% accuracy based on image.

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

![image/png](https://cdn-uploads.huggingface.co/production/uploads/6449300e3adf50d864095b90/QG0rlWj7WTEPnci8qdvFg.png)

```
Classification report:

              precision    recall  f1-score   support

     battery     0.9466    0.9484    0.9475      2130
  biological     0.9725    0.9812    0.9769      2130
 brown-glass     0.9609    0.9690    0.9649      2130
   cardboard     0.9920    0.9347    0.9625      2130
     clothes     0.9469    0.9789    0.9626      2130
 green-glass     0.9231    0.9869    0.9539      2130
       metal     0.8810    0.9038    0.8922      2130
       paper     0.9215    0.9427    0.9320      2130
     plastic     0.8860    0.6643    0.7593      2130
       shoes     0.9383    0.9854    0.9613      2130
       trash     0.8643    0.9718    0.9149      2130
 white-glass     0.9049    0.8709    0.8876      2130

    accuracy                         0.9282     25560
   macro avg     0.9282    0.9282    0.9263     25560
weighted avg     0.9282    0.9282    0.9263     25560
```