text-generation-webui/extensions/send_pictures/script.py
2023-04-05 13:54:50 -03:00

47 lines
2.1 KiB
Python

import base64
from io import BytesIO
import gradio as gr
import modules.chat as chat
import modules.shared as shared
import torch
from PIL import Image
from transformers import BlipForConditionalGeneration, BlipProcessor
# If 'state' is True, will hijack the next chat generation with
# custom input text given by 'value' in the format [text, visible_text]
input_hijack = {
'state': False,
'value': ["", ""]
}
processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base", torch_dtype=torch.float32).to("cpu")
def caption_image(raw_image):
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)
def generate_chat_picture(picture, name1, name2):
text = f'*{name1} sends {name2} a picture that contains the following: "{caption_image(picture)}"*'
# lower the resolution of sent images for the chat, otherwise the log size gets out of control quickly with all the base64 values in visible history
picture.thumbnail((300, 300))
buffer = BytesIO()
picture.save(buffer, format="JPEG")
img_str = base64.b64encode(buffer.getvalue()).decode('utf-8')
visible_text = f'<img src="data:image/jpeg;base64,{img_str}" alt="{text}">'
return text, visible_text
def ui():
picture_select = gr.Image(label='Send a picture', type='pil')
# Prepare the hijack with custom inputs
picture_select.upload(lambda picture, name1, name2: input_hijack.update({"state": True, "value": generate_chat_picture(picture, name1, name2)}), [picture_select, shared.gradio['name1'], shared.gradio['name2']], None)
# Call the generation function
picture_select.upload(chat.cai_chatbot_wrapper, shared.input_params, shared.gradio['display'], show_progress=shared.args.no_stream)
# Clear the picture from the upload field
picture_select.upload(lambda : None, [], [picture_select], show_progress=False)