Views
No views yet
huggingface-cli download wjpoom/SPEC-CLIP-ViT-B-32 --local-dir checkpoints/SPEC-CLIP-ViT-B-321# pip install open_clip_torch
2import torch
3from PIL import Image
4import open_clip
5
6model, _, preprocess = open_clip.create_model_and_transforms('ViT-B-32', pretrained='checkpoints/SPEC-CLIP-ViT-B-32', load_weights_only=False)
7model.eval() # model in train mode by default, impacts some models with BatchNorm or stochastic depth active
8tokenizer = open_clip.get_tokenizer('ViT-B-32')
9
10image = preprocess(Image.open("docs/CLIP.png")).unsqueeze(0)
11text = tokenizer(["a diagram", "a dog", "a cat"])
12
13with torch.no_grad(), torch.autocast("cuda"):
14 image_features = model.encode_image(image)
15 text_features = model.encode_text(text)
16 image_features /= image_features.norm(dim=-1, keepdim=True)
17 text_features /= text_features.norm(dim=-1, keepdim=True)
18
19 text_probs = (100.0 * image_features @ text_features.T).softmax(dim=-1)
20
21print("Label probs:", text_probs) # prints: [[1., 0., 0.]]1@inproceedings{peng2024synthesize,
2 title={Synthesize diagnose and optimize: Towards fine-grained vision-language understanding},
3 author={Peng, Wujian and Xie, Sicheng and You, Zuyao and Lan, Shiyi and Wu, Zuxuan},
4 booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
5 pages={13279--13288},
6 year={2024}
7}