-
Upload folder using huggingface_hub
Fullname
-
Upload folder using huggingface_hub
Fullname
-
Upload folder using huggingface_hub
Fullname
-
Update README.md
Fullname
-
Update README.md
Fullname
-
Update README.md
Fullname
-
Update README.md
Fullname
# Pokemon Team Classification with Vision Transformer
A fine-tuned Vision Transformer (ViT) model for classifying 6 specific Pokemon from a competitive team setup. This model can identify Arceus, Marshadow, Sandy Shocks, Slaking, Reshiram, and Magearna with high accuracy.
## Model Details
- **Base Model**: `google/vit-base-patch16-224`
- **Model Type**: Vision Transformer for Image Classification
- **Classes**: 6 Pokemon (Arceus, Marshadow, Sandy Shocks, Slaking, Reshiram, Magearna)
- **Input Size**: 224x224 RGB images
- **Framework**: PyTorch + Transformers
## Training Details
### Dataset
- **Arceus**: 644 images
- **Marshadow**: 101 images
- **Sandy Shocks**: 75 images
- **Slaking**: 152 images
- **Reshiram**: 118 images
- **Magearna**: 200 images
### Training Strategy
- **Balanced Sampling**: Each epoch uses exactly 75 samples per class to prevent overfitting on Arceus
- **Data Augmentation**: Random horizontal flip, rotation (±15°), color jitter, and resized crop
- **Transf… -
Upload folder using huggingface_hub
Fullname
-
initial commit
Fullname
