gradio app

app.py

@@ -2,6 +2,9 @@ from transformers import InstructBlipProcessor, InstructBlipForConditionalGenera
 from PIL import Image
 import torch
 
+import gradio as gr
+
+
 device = "cuda" if torch.cuda.is_available() else "cpu"
 print(f"Using device: {device}")
 
@@ -14,22 +17,22 @@ model = InstructBlipForConditionalGeneration.from_pretrained(
     device_map="auto"
 )
 
-# Load your image
-image = Image.open("example.jpg").convert("RGB")
-
 # Prompt to force paragraph-level description
 prompt = (
     "Describe this image in a detailed paragraph of 5-7 sentences. "
     "Mention setting, objects, colors, actions, background details, and possible context."
 )
 
-inputs = processor(images=image, text=prompt, return_tensors="pt").to(device)
+def caption_image(image):
+    inputs = processor(images=image, text=prompt, return_tensors="pt").to(device)
+    out = model.generate(**inputs, max_new_tokens=250, temperature=0.7, top_p=0.9)
+    return processor.batch_decode(out, skip_special_tokens=True)[0]
 
-out = model.generate(
-    **inputs,
-    max_new_tokens=250,  # enough for multi-sentence
-    temperature=0.7,
-    top_p=0.9
+demo = gr.Interface(
+    fn=caption_image, 
+    inputs=gr.Image(type="pil"), 
+    outputs="text",
+    title="Image to Paragraph Captioning"
 )
 
-print(processor.batch_decode(out, skip_special_tokens=True)[0])
+demo.launch()