|
|
from transformers import ViTFeatureExtractor, ViTForImageClassification |
|
|
from PIL import Image |
|
|
import requests |
|
|
|
|
|
url = 'http: |
|
|
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 |
|
|
# model predicts one of the 1000 ImageNet classes |
|
|
predicted_class_idx = logits.argmax(-1).item() |
|
|
print("Predicted class:", model.config.id2label[predicted_class_idx |
