Spaces:
Sleeping
Sleeping
from PIL import Image | |
from transformers import BlipProcessor, BlipForConditionalGeneration | |
import gradio as gr | |
# Initialization of the BLIP processor and model | |
processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-large") | |
model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large") | |
def generate_captions(image, text=""): | |
# Convert the uploaded image to PIL Image | |
raw_image = Image.fromarray(image).convert('RGB') | |
if text: # Conditional image captioning | |
inputs = processor(raw_image, text, return_tensors="pt") | |
else: # Unconditional image captioning | |
inputs = processor(raw_image, return_tensors="pt") | |
# Generate captions for the image | |
out = model.generate(**inputs) | |
caption = processor.decode(out[0], skip_special_tokens=True) | |
return caption | |
# Gradio Interface | |
iface = gr.Interface( | |
fn=generate_captions, | |
inputs=[ | |
gr.Image(label="Upload/Drag Image"), # Removed the 'tool' argument | |
gr.Textbox(label="Conditional Text (optional)", placeholder="Enter conditional text (optional)...") | |
], | |
outputs=gr.Textbox(label="Generated Caption"), | |
title="BLIP Image Caption Generator", | |
description="This app generates captions for uploaded images. You can also provide conditional text to guide the caption generation." | |
) | |
if __name__ == "__main__": | |
iface.launch() | |