add async to speed up 2x

example5_async.py

@@ -0,0 +1,49 @@
+"""
+Example 5: Async/Pipelined Processing
+
+Overlaps image capture with deformation computation for higher throughput.
+Pipeline: [Capture Thread] -> [Queue] -> [Compute Thread] -> [Queue] -> [Main]
+"""
+
+import os
+os.environ["OPENCV_VIDEOIO_MSMF_ENABLE_HW_TRANSFORMS"] = "0"
+
+import cv2
+import orisys
+
+
+def main():
+    sensor = orisys.Sensor("./s1.mp4", isstitch=True, cuda=True, verbose=True)
+    sensor.enable_async(queue_size=2)
+    
+    print("\nPress 'q' to quit, 'r' to reset")
+    
+    try:
+        while True:
+            data = sensor.get_async(timeout=0.05)
+            if data is None:
+                continue
+            
+            print(f"FPS: {data['fps']:5.1f}, Fn: {data['fnormal']:8.2f}, "
+                  f"Fx: {data['fshearx']:8.2f}, Fy: {data['fsheary']:8.2f}")
+            
+            if data['img'] is not None and data['flow'] is not None:
+                arrows = orisys.util.draw_arrows(data['img'].copy(), data['flow'], threshold=5)
+                cv2.imshow("Deformation", arrows)
+            
+            if data['depth'] is not None:
+                depth_img = cv2.applyColorMap((data['depth'] * 255).astype('uint8'), cv2.COLORMAP_HOT)
+                cv2.imshow("Depth", depth_img)
+            
+            key = cv2.waitKey(1) & 0xFF
+            if key == ord('q'):
+                break
+            elif key == ord('r'):
+                sensor.reset()
+    finally:
+        sensor.disconnect()
+        cv2.destroyAllWindows()
+
+
+if __name__ == '__main__':
+    main()