Create app.py

app.py

@@ -0,0 +1,32 @@
+# app.py
+import gradio as gr
+import cv2, os, pandas as pd
+from inference_utils import load_models, process_frame
+
+vehicle_detector, vehicle_classifier, plate_detector, ocr_reader = load_models()
+log_path = "vehicle_log.csv"
+if not os.path.exists(log_path):
+    pd.DataFrame(columns=["timestamp", "vehicle_type", "plate_text"]).to_csv(log_path, index=False)
+
+def run_pipeline(video):
+    cap = cv2.VideoCapture(video)
+    logs = []
+
+    while cap.isOpened():
+        ret, frame = cap.read()
+        if not ret: break
+        frame = cv2.resize(frame, (640, 480))
+        res = process_frame(frame, vehicle_detector, vehicle_classifier, plate_detector, ocr_reader)
+        logs.extend(res)
+
+    df = pd.DataFrame(logs, columns=["timestamp", "vehicle_type", "plate_text"])
+    df.to_csv(log_path, mode='a', header=False, index=False)
+    return f"Processed {len(logs)} vehicles. Log saved to {log_path}."
+
+iface = gr.Interface(fn=run_pipeline,
+                     inputs=gr.Video(label="Upload Video"),
+                     outputs="text",
+                     title="Vehicle ANPR Pipeline",
+                     description="Uploads a video, detects vehicles, classifies type, reads plate and logs to CSV.")
+
+iface.launch()