Spaces:
Sleeping
Sleeping
File size: 1,595 Bytes
e8994f0 | 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 | '''
import torch
from torchvision import models, transforms
from PIL import Image
import requests
from io import BytesIO
# Load ResNet50 model
model = models.resnet50(pretrained=True)
model.eval()
preprocess = transforms.Compose([
transforms.Resize(256),
transforms.CenterCrop(224),
transforms.ToTensor(),
])
def get_embedding(image: Image.Image):
img_tensor = preprocess(image).unsqueeze(0)
with torch.no_grad():
embedding = model(img_tensor).squeeze().numpy()
return embedding
def load_image_from_url(url):
response = requests.get(url)
img = Image.open(BytesIO(response.content)).convert("RGB")
return img
'''
import torch
import open_clip
from PIL import Image
import requests
from io import BytesIO
from torchvision import transforms
# Load CLIP model (ViT-B-32 is a good default)
model, _, preprocess = open_clip.create_model_and_transforms(
'ViT-B-32', pretrained='openai'
)
model.eval()
# Define a function to compute CLIP embeddings
def get_embedding(image: Image.Image):
# CLIP expects images in a certain format:
image_input = preprocess(image).unsqueeze(0) # (1, 3, H, W)
with torch.no_grad():
image_features = model.encode_image(image_input)
# Normalize to unit vector for cosine similarity
image_features = image_features / image_features.norm(dim=-1, keepdim=True)
return image_features.squeeze().cpu().numpy()
# Helper to load image from URL
def load_image_from_url(url):
response = requests.get(url)
img = Image.open(BytesIO(response.content)).convert("RGB")
return img
|