|
--- |
|
license: apache-2.0 |
|
tags: |
|
- vision |
|
- image-classification |
|
widget: |
|
- src: https://datasets-server.huggingface.co/assets/ShihTing/IsCausewayOffset/--/ShihTing--IsCausewayOffset/validation/0/image/image.jpg |
|
example_title: Ex1 |
|
--- |
|
# PanJu offset detect by image |
|
Use fintune from google/vit-base-patch16-224(https://huggingface.co/google/vit-base-patch16-224) |
|
|
|
## Dataset |
|
```python |
|
DatasetDict({ |
|
train: Dataset({ |
|
features: ['image', 'label'], |
|
num_rows: 329 |
|
}) |
|
validation: Dataset({ |
|
features: ['image', 'label'], |
|
num_rows: 56 |
|
}) |
|
}) |
|
|
|
``` |
|
36 Break and 293 Normal in train |
|
5 Break and 51 Normal in validation |
|
|
|
|
|
## Intended uses |
|
|
|
### How to use |
|
|
|
Here is how to use this model to classify an image of the COCO 2017 dataset into one of the 1,000 ImageNet classes: |
|
|
|
```python |
|
# Load image |
|
import torch |
|
from transformers import ViTFeatureExtractor, ViTForImageClassification,AutoModel |
|
from PIL import Image |
|
import requests |
|
url='https://datasets-server.huggingface.co/assets/ShihTing/IsCausewayOffset/--/ShihTing--IsCausewayOffset/validation/0/image/image.jpg' |
|
image = Image.open(requests.get(url, stream=True).raw) |
|
# Load model |
|
from transformers import AutoFeatureExtractor, AutoModelForImageClassification |
|
device = torch.device('cpu') |
|
extractor = AutoFeatureExtractor.from_pretrained('ShihTing/PanJuOffset_TwoClass') |
|
model = AutoModelForImageClassification.from_pretrained('ShihTing/PanJuOffset_TwoClass') |
|
# Predict |
|
inputs = extractor(images=image, return_tensors="pt") |
|
outputs = model(**inputs) |
|
logits = outputs.logits |
|
Prob = outputs.logits.softmax(dim=-1).tolist() |
|
print(Prob) |
|
# model predicts one of the 1000 ImageNet classes |
|
predicted_class_idx = logits.argmax(-1).item() |
|
print("Predicted class:", model.config.id2label[predicted_class_idx]) |
|
``` |
|
|
|
|