Spaces:
Running
Running
File size: 430 Bytes
73d7797 |
1 2 3 4 5 6 7 8 9 10 11 |
from transformers import AutoProcessor, BlipForQuestionAnswering
vqa_id = "Salesforce/blip-vqa-base"
vqa_model = BlipForQuestionAnswering.from_pretrained(vqa_id)
vqa_processor = AutoProcessor.from_pretrained(vqa_id)
def visual_qa(image, question):
inputs = vqa_processor(image, question, return_tensors="pt")
out = vqa_model.generate(**inputs)
return vqa_processor.decode(out[0], skip_special_tokens=True)
|