Views
No views yet
1import requests
2from PIL import Image
3from transformers import BlipProcessor, BlipForConditionalGeneration
4
5processor = BlipProcessor.from_pretrained(""moranyanuka/blip-image-captioning-base-mocha"")
6model = BlipForConditionalGeneration.from_pretrained("moranyanuka/blip-image-captioning-base-mocha")
7
8img_url = 'https://storage.googleapis.com/sfr-vision-language-research/BLIP/demo.jpg'
9raw_image = Image.open(requests.get(img_url, stream=True).raw).convert('RGB')
10
11# conditional image captioning
12text = "a photography of"
13inputs = processor(raw_image, text, return_tensors="pt")
14
15out = model.generate(**inputs)
16print(processor.decode(out[0], skip_special_tokens=True))
17
18# unconditional image captioning
19inputs = processor(raw_image, return_tensors="pt")
20
21out = model.generate(**inputs)
22print(processor.decode(out[0], skip_special_tokens=True))1import requests
2from PIL import Image
3from transformers import BlipProcessor, BlipForConditionalGeneration
4
5processor = BlipProcessor.from_pretrained("moranyanuka/blip-image-captioning-base-mocha")
6model = BlipForConditionalGeneration.from_pretrained("moranyanuka/blip-image-captioning-base-mocha").to("cuda")
7
8img_url = 'https://storage.googleapis.com/sfr-vision-language-research/BLIP/demo.jpg'
9raw_image = Image.open(requests.get(img_url, stream=True).raw).convert('RGB')
10
11# conditional image captioning
12text = "a photography of"
13inputs = processor(raw_image, text, return_tensors="pt").to("cuda")
14
15out = model.generate(**inputs)
16print(processor.decode(out[0], skip_special_tokens=True))
17
18# unconditional image captioning
19inputs = processor(raw_image, return_tensors="pt").to("cuda")
20
21out = model.generate(**inputs)
22print(processor.decode(out[0], skip_special_tokens=True))float16)1import torch
2import requests
3from PIL import Image
4from transformers import BlipProcessor, BlipForConditionalGeneration
5
6processor = BlipProcessor.from_pretrained("moranyanuka/blip-image-captioning-base-mocha")
7model = BlipForConditionalGeneration.from_pretrained("moranyanuka/blip-image-captioning-base-mocha", torch_dtype=torch.float16).to("cuda")
8
9img_url = 'https://storage.googleapis.com/sfr-vision-language-research/BLIP/demo.jpg'
10raw_image = Image.open(requests.get(img_url, stream=True).raw).convert('RGB')
11
12# conditional image captioning
13text = "a photography of"
14inputs = processor(raw_image, text, return_tensors="pt").to("cuda", torch.float16)
15
16out = model.generate(**inputs)
17print(processor.decode(out[0], skip_special_tokens=True))
18# >>> a photography of a woman and her dog on the beach
19
20# unconditional image captioning
21inputs = processor(raw_image, return_tensors="pt").to("cuda", torch.float16)
22
23out = model.generate(**inputs)
24print(processor.decode(out[0], skip_special_tokens=True))
25>>> a woman sitting on the beach with a dog@misc{benkish2024mitigating,
title={Mitigating Open-Vocabulary Caption Hallucinations},
author={Assaf Ben-Kish and Moran Yanuka and Morris Alper and Raja Giryes and Hadar Averbuch-Elor},
year={2024},
eprint={2312.03631},
archivePrefix={arXiv},
primaryClass={cs.CV}
}