|
--- |
|
tags: |
|
- fastai |
|
- resnet |
|
- computer-vision |
|
- classification |
|
- binary-classification |
|
license: |
|
- cc0-1.0 |
|
--- |
|
|
|
# Resnet34 Pokemon Card Classifier |
|
|
|
## Model Description |
|
|
|
This is a resnet34 model fine-tuned with fastai to [classify real and fake Pokemon cards (dataset)](https://www.kaggle.com/datasets/ongshujian/real-and-fake-pokemon-cards). |
|
|
|
Here is a colab notebook that shows how the model was trained and pushed to the hub: [link](https://github.com/mindwrapped/pokemon-card-checker/blob/main/pokemon_card_checker.ipynb). |
|
|
|
## Intended uses & limitation |
|
|
|
This model is trained to identify real vs fake cards based on the backs of the cards, not the front. |
|
|
|
## How to use |
|
|
|
```python |
|
from huggingface_hub import from_pretrained_fastai |
|
|
|
# Pull model from hub |
|
learn = from_pretrained_fastai('hugginglearners/pokemon-card-checker') |
|
|
|
# Get prediction for this image |
|
pred_label, _, scores = learn.predict(img) |
|
``` |
|
|
|
## Training data |
|
|
|
Dataset located here: [link](https://www.kaggle.com/datasets/ongshujian/real-and-fake-pokemon-cards). |
|
|
