colab上比较clip

!pip install torch torchvision ftfy regex
!pip install git+https://github.com/openai/CLIP.git

import torch
import clip
from PIL import Image
from io import BytesIO
import requests

# 加载模型
device = "cuda" if torch.cuda.is_available() else "cpu"
model, preprocess = clip.load("ViT-B/32", device=device)

# 加载并预处理图片
image1 = preprocess(Image.open('img1.jpg')).unsqueeze(0).to(device)
image2 = preprocess(Image.open('img2.jpg')).unsqueeze(0).to(device)

with torch.no_grad():
    image_features1 = model.encode_image(image1)
    image_features2 = model.encode_image(image2)

# 计算余弦相似度
cosine_similarity = torch.nn.CosineSimilarity(dim=1, eps=1e-6)
similarity = cosine_similarity(image_features1, image_features2)
print("Similarity score:", similarity.item())