text-generation-webui/modules/bot_picture.py

11 lines
540 B
Python
Raw Normal View History

import torch
2023-02-23 18:41:42 +01:00
from transformers import BlipForConditionalGeneration, BlipProcessor
2023-02-15 00:38:21 +01:00
processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
2023-02-15 04:03:19 +01:00
model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base", torch_dtype=torch.float32).to("cpu")
def caption_image(raw_image):
2023-02-15 04:03:19 +01:00
inputs = processor(raw_image.convert('RGB'), return_tensors="pt").to("cpu", torch.float32)
out = model.generate(**inputs, max_new_tokens=100)
return processor.decode(out[0], skip_special_tokens=True)