event driven system

test_tracking_realtime.py

@@ -5,11 +5,10 @@ This script demonstrates:
 - Event-driven stream processing with StreamConnectionManager
 - Batched GPU inference with ModelController
 - Ping-pong buffer architecture for optimal throughput
-- Async/await pattern for multiple RTSP streams
+- Callback-based event-driven pattern for RTSP streams
 - Automatic PT to TensorRT conversion
 """
 
-import asyncio
 import time
 import os
 import torch
@@ -26,7 +25,7 @@ from services import (
 load_dotenv()
 
 
-async def main_single_stream():
+def main_single_stream():
     """Single stream example with event-driven architecture."""
     print("=" * 80)
     print("Event-Driven GPU-Accelerated Object Tracking - Single Stream")
@@ -64,9 +63,9 @@ async def main_single_stream():
     print("\n[2/3] Initializing with PT model...")
     print("Note: First load will convert PT to TensorRT (3-5 minutes)")
     print("Subsequent loads will use cached TensorRT engine\n")
-    
+
     try:
-        await manager.initialize(
+        manager.initialize(
             model_path=MODEL_PATH,
             model_id="detector",
             preprocess_fn=YOLOv8Utils.preprocess,
@@ -85,7 +84,7 @@ async def main_single_stream():
     # Connect stream
     print("\n[3/3] Connecting to stream...")
     try:
-        connection = await manager.connect_stream(
+        connection = manager.connect_stream(
             rtsp_url=STREAM_URL,
             stream_id="camera_1",
             buffer_size=30
@@ -110,7 +109,7 @@ async def main_single_stream():
         cv2.resizeWindow("Object Tracking", 1280, 720)
 
     try:
-        async for result in connection.tracking_results():
+        for result in connection.tracking_results():
             result_count += 1
 
             # Check if we've reached max frames
@@ -189,8 +188,8 @@ async def main_single_stream():
     if ENABLE_DISPLAY:
         cv2.destroyAllWindows()
 
-    await connection.stop()
-    await manager.shutdown()
+    connection.stop()
+    manager.shutdown()
     print("✓ Stopped")
 
     # Final stats
@@ -199,7 +198,7 @@ async def main_single_stream():
     print(f"\nFinal: {result_count} results in {elapsed:.1f}s ({avg_fps:.1f} FPS)")
 
 
-async def main_multi_stream():
+def main_multi_stream():
     """Multi-stream example with batched inference."""
     print("=" * 80)
     print("Event-Driven GPU-Accelerated Object Tracking - Multi-Stream")
@@ -245,7 +244,7 @@ async def main_multi_stream():
     # Initialize with PT model
     print("\n[2/3] Initializing with PT model...")
     try:
-        await manager.initialize(
+        manager.initialize(
             model_path=MODEL_PATH,
             model_id="detector",
             preprocess_fn=YOLOv8Utils.preprocess,
@@ -266,7 +265,7 @@ async def main_multi_stream():
     connections = {}
     for stream_id, rtsp_url in camera_urls:
         try:
-            conn = await manager.connect_stream(
+            conn = manager.connect_stream(
                 rtsp_url=rtsp_url,
                 stream_id=stream_id,
                 buffer_size=30
@@ -295,7 +294,7 @@ async def main_multi_stream():
         # Simple approach: iterate over first connection's results
         # In production, you'd properly merge all result streams
         for conn in connections.values():
-            async for result in conn.tracking_results():
+            for result in conn.tracking_results():
                 total_results += 1
                 stream_id = result.stream_id
 
@@ -322,8 +321,8 @@ async def main_multi_stream():
     print(f"{'=' * 80}")
 
     for conn in connections.values():
-        await conn.stop()
-    await manager.shutdown()
+        conn.stop()
+    manager.shutdown()
     print("✓ Stopped")
 
     # Final stats
@@ -335,6 +334,6 @@ async def main_multi_stream():
 if __name__ == "__main__":
     import sys
     if len(sys.argv) > 1 and sys.argv[1] == "single":
-        asyncio.run(main_single_stream())
+        main_single_stream()
     else:
-        asyncio.run(main_multi_stream())
+        main_multi_stream()