File size: 1,541 Bytes
3cf7762 6e6df1d 3cf7762 6e6df1d 3cf7762 6e6df1d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 | ---
language: en
license: mit
library_name: fastai
pipeline_tag: image-classification
tags:
- fastai
- timm
- vision-transformer
- image-classification
- pc-parts
widget:
- src: https://images.unsplash.com/photo-1587202372775-a457f4ad61b9
example_title: PC build example
---
# vit_base_patch16_pc_parts_classifier
Vision Transformer image classifier for 11 PC component and cable-management classes.
## Model Details
- Architecture: ViT-Base Patch16 224 (`vit_base_patch16_224`)
- Framework: FastAI + timm + PyTorch
- Input size: 224x224 RGB
- Classes: 11
- Epochs: 15
- Batch size: 16
- Test accuracy: 0.7389
- Training date: 2026-04-17
## Labels
1. AIO_Liquid_Cooler
2. Air_Cooler
3. Bad_Cable_Management
4. CPU
5. Good_Cable_Management
6. Graphics_Card
7. M2_NVMe_Drive
8. Motherboard
9. PC_Case
10. Power_Supply
11. RAM_Stick
## Files
- `best_model_export.pkl`: FastAI export for direct inference.
- `best_model_state_dict.pth`: PyTorch state dict.
- `best_model_metadata.json`: Training and class metadata.
## Inference (FastAI)
```python
from fastai.learner import load_learner
from pathlib import Path
learn = load_learner("best_model_export.pkl")
pred_class, pred_idx, probs = learn.predict(Path("sample.jpg"))
print(pred_class)
print({learn.dls.vocab[i]: float(probs[i]) for i in range(len(probs))})
```
## Live Demo
A live inference demo is available on Hugging Face Spaces:
- https://huggingface.co/spaces/shaanzeeeee/vit-base-pc-parts-inference
|