app.py

import torch
import requests
from PIL import Image
from transformers import BlipProcessor, BlipForConditionalGeneration
import gradio as gr
device="cpu"
processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base", torch_dtype=torch.float16).to(device)

# Function to process the image and generate captions
def generate_caption(image, caption_type, text):
    raw_image = Image.fromarray(image.astype('uint8'), 'RGB')
    
    if caption_type == "Conditional":
        caption = conditional_image_captioning(raw_image, text)
    else:
        caption = unconditional_image_captioning(raw_image)
    
    return caption

# Conditional image captioning
def conditional_image_captioning(raw_image, text):
    inputs = processor(raw_image, text, return_tensors="pt").to(device, torch.float16)
    out = model.generate(**inputs)
    caption = processor.decode(out[0], skip_special_tokens=True)
    return caption

# Unconditional image captioning
def unconditional_image_captioning(raw_image):
    inputs = processor(raw_image, return_tensors="pt").to(device, torch.float16)
    out = model.generate(**inputs)
    caption = processor.decode(out[0], skip_special_tokens=True)
    return caption

# Interface setup
input_image = gr.inputs.Image()
input_text = gr.inputs.Textbox(label="Enter Text (for Conditional Captioning)")

choices = ["Conditional", "Unconditional"]
radio_button = gr.inputs.Radio(choices, label="Captioning Type")

output_text = gr.outputs.Textbox(label="Caption")

# Create the interface
gr.Interface(fn=generate_caption, inputs=[input_image, radio_button, input_text], outputs=output_text, title="Image Captioning",debug=True).launch()