Create app.py

app.py

@@ -0,0 +1,71 @@
+import torch
+from PIL import Image
+from transformers import AutoProcessor, AutoModelForCausalLM
+import gradio as gr
+import json
+import traceback
+import os
+
+model_name = "meta-llama/Llama-3.2-11B-Vision-Instruct"
+token = os.environ.get("HUGGINGFACE_TOKEN")
+processor = AutoProcessor.from_pretrained(model_name, token=token)
+model = AutoModelForCausalLM.from_pretrained(
+    model_name,
+    quantization_config={"load_in_4bit": True},
+    token=token
+)
+
+if torch.cuda.is_available():
+    model = model.to('cuda')
+
+def analyze_image(image, prompt):
+    try:
+        messages = [
+            {"role": "user", "content": [
+                {"type": "image"},
+                {"type": "text", "text": prompt}
+            ]}
+        ]
+        input_text = processor.apply_chat_template(messages, add_generation_prompt=True)
+        inputs = processor(
+            images=image,
+            text=input_text,
+            return_tensors="pt"
+        ).to(model.device)
+        with torch.no_grad():
+            output = model.generate(**inputs, max_new_tokens=100)
+        result = processor.decode(output[0], skip_special_tokens=True)
+        try:
+            return json.loads(result)
+        except json.JSONDecodeError:
+            return {"error": "Failed to parse model output as JSON", "raw_output": result}
+    except Exception as e:
+        return {"error": str(e), "traceback": traceback.format_exc()}
+
+default_prompt = """Analyze this image and determine if it contains a data logger. 
+A data logger is typically a small, black electronic device used to monitor and record data 
+over time, such as voltage, temperature, or current, via external sensors.
+
+If a data logger is present in the image, respond with:
+{"present": true, "reason": "Brief explanation of why you believe it's a data logger"}
+
+If no data logger is visible, respond with:
+{"present": false, "reason": "Brief explanation of why you believe there's no data logger"}
+
+Ensure your response is in valid JSON format."""
+
+iface = gr.Interface(
+    fn=analyze_image,
+    inputs=[
+        gr.Image(type="pil", label="Upload Image"),
+        gr.Textbox(label="Prompt", default=default_prompt, lines=10)
+    ],
+    outputs=gr.JSON(label="Analysis Result"),
+    title="Data Logger Detection using Llama 3.2 Vision",
+    description="Upload an image and customize the prompt to check if it contains a data logger.",
+    examples=[
+        ["bad.jpg", default_prompt]
+    ]
+)
+
+iface.launch()