Skip to content

Instantly share code, notes, and snippets.

@maziyarpanahi

maziyarpanahi/hf-vit-pytorch.py

Last active August 25, 2022 13:48
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save maziyarpanahi/64f0575cdccd383242c6683c0def46cb to your computer and use it in GitHub Desktop.
Save maziyarpanahi/64f0575cdccd383242c6683c0def46cb to your computer and use it in GitHub Desktop.
Download ZIP
Raw
hf-vit-pytorch.py
from transformers import ViTFeatureExtractor, ViTForImageClassification
from PIL import Image
import requests
url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
image = Image.open(requests.get(url, stream=True).raw)
feature_extractor = ViTFeatureExtractor.from_pretrained('google/vit-base-patch16-224')
model = ViTForImageClassification.from_pretrained('google/vit-base-patch16-224')
inputs = feature_extractor(images=image, return_tensors="pt")
outputs = model(**inputs)
logits = outputs.logits
predicted_class_idx = logits.argmax(-1).item()
print("Predicted class:", model.config.id2label[predicted_class_idx])
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment