Update app.py

app.py

@@ -1,48 +1,76 @@
 import gradio as gr
-from inference_sdk import InferenceHTTPClient
-from PIL import Image
-import requests
-
-# Initialize the Roboflow client
-client = InferenceHTTPClient(
-    api_url="https://detect.roboflow.com",
-    api_key="dxkgGGHSZ3DI8XzVn29U"
-)
-
-def predict(image: Image.Image):
-    try:
-        # Save image temporarily
-        image.save("temp.jpg")
-        with open("temp.jpg", "rb") as f:
-            result = client.run_workflow(
-                workspace_name="naveen-kumar-hnmil",
-                workflow_id="detect-count-and-visualize-5",
-                images={"image": f},
-                use_cache=True
-            )
-        
-        # Extract annotated image
-        annotated_url = result.get("visualizations", {}).get("image")
-        if not annotated_url:
-            return "Error: No visualization returned from Roboflow."
-        
-        response = requests.get(annotated_url, stream=True)
-        if response.status_code != 200:
-            return f"Error fetching image from: {annotated_url}"
-        
-        return Image.open(response.raw)
-
-    except Exception as e:
-        return f"Exception: {str(e)}"
-
-# Gradio UI
-demo = gr.Interface(
-    fn=predict,
-    inputs=gr.Image(type="pil"),
-    outputs="image",
-    title="Solar Panel Fault Detection",
-    description="Upload an image and get predictions from your Roboflow workflow."
-)
+import cv2
+import numpy as np
+import tempfile
+import os
+import pandas as pd
+from inference import InferencePipeline
 
-demo.launch()
+def run_inference(video_file):
+    if video_file is None:
+        return None, None, "Please upload a video file."
+
+    temp_video_path = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4").name
+    with open(temp_video_path, "wb") as f:
+        f.write(video_file.read())
+
+    annotated_frames = []
+    prediction_log = []
+
+    def on_prediction(result, frame):
+        if result.get("output_image") is not None:
+            image_np = result["output_image"].numpy_image
+            annotated_frames.append(image_np)
+
+        if result.get("predictions"):
+            for pred in result["predictions"].get("predictions", []):
+                prediction_log.append({
+                    "Class": pred["class"],
+                    "Confidence": round(pred["confidence"], 2),
+                    "Box": f'({pred["x"]}, {pred["y"]}, {pred["width"]}, {pred["height"]})'
+                })
+
+    pipeline = InferencePipeline.init_with_workflow(
+        api_key="dxkgGGHSZ3DI8XzVn29U",
+        workspace_name="naveen-kumar-hnmil",
+        workflow_id="detect-count-and-visualize-8",
+        video_reference=temp_video_path,
+        max_fps=5,
+        on_prediction=on_prediction
+    )
+
+    pipeline.start()
+    pipeline.join()
+
+    os.unlink(temp_video_path)
 
+    if not annotated_frames:
+        return None, None, "No faults detected or output frames returned."
+
+    # Save annotated video
+    height, width, _ = annotated_frames[0].shape
+    output_path = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4").name
+    out = cv2.VideoWriter(output_path, cv2.VideoWriter_fourcc(*"mp4v"), 5, (width, height))
+    for frame in annotated_frames:
+        out.write(cv2.cvtColor(frame, cv2.COLOR_RGB2BGR))
+    out.release()
+
+    video_bytes = open(output_path, "rb").read()
+    os.unlink(output_path)
+
+    # Prepare dataframe as CSV text for Gradio table
+    df = pd.DataFrame(prediction_log)
+
+    return video_bytes, df, "Inference completed successfully."
+
+with gr.Blocks() as demo:
+    gr.Markdown("# Solar Panel Fault Detection (Roboflow + Gradio)")
+    video_input = gr.Video(label="Upload Thermal Video", source="upload", type="file")
+    output_video = gr.Video(label="Annotated Output Video")
+    output_table = gr.Dataframe(headers=["Class", "Confidence", "Box"], interactive=False)
+    status = gr.Textbox(label="Status", interactive=False)
+
+    run_btn = gr.Button("Run Inference")
+    run_btn.click(fn=run_inference, inputs=video_input, outputs=[output_video, output_table, status])
+
+demo.launch()