Skip to main content

Preview CSI Camera Image

This section explains how to preview live images from a CSI camera (such as a Raspberry Pi CSI camera) using OpenCV and jetcam on NVIDIA Jetson platforms.

Previewing a CSI camera is commonly used for: - Verifying CSI camera connectivity - Real-time image inspection - Jetson-based vision development and debugging

1. Implementation Principle​

Use the CSICamera() class from the jetcam.csi_camera module to capture images from a CSI camera.

The CSI camera provides frames directly without using /dev/video* in the same way as USB cameras.

2. Implementation Effect​

Navigate to the OpenCV working directory:

cd ~/opencv

Run the CSI camera preview script:

python3 18.camera_preview_csi.py
note

The program opens CSI camera 0 by default. If multiple CSI cameras are available, modify the capture_device index in the code.

Select the camera preview window and press q to exit the program.

CSI Camera Preview Result

3. Implementation Code​

import cv2
from jetcam.csi_camera import CSICamera

def preview_csi_camera():
    cap = CSICamera(
        capture_device=0,
        width=640,
        height=480
    )

    while True:
        frame = cap.read()

        if frame is not None:
            cv2.imshow(
                'CSI Camera Preview',
                cv2.resize(frame, (640, 480))
            )

            if cv2.waitKey(1) & 0xFF == ord('q'):
                break
        else:
            print("Error: Could not open CSI camera.")
            break

    cap.release()
    cv2.destroyAllWindows()

preview_csi_camera()

4. Code Explanation​

  • CSICamera(capture_device=0, width=640, height=480) initializes the CSI camera\
  • cap.read() captures a frame\
  • cv2.imshow() displays the live feed\
  • cv2.waitKey(1) listens for keyboard input\
  • cap.release() releases the camera

Maintained by HemiHex for Jetson and OpenCV-based computer vision workflows.