An implementation of OpenDataCam 2.0.1 that is fully containerized and runs on balena + Jetson Nano.
A more detailed tutorial will be available soon.
This project now uses our experimental jetson-nano-cuda-cudnn-opencv image, so you no longer need to download the NVIDIA SDK Manager. You can use the deploy button below to create and build the application in your balenaCloud account. (balenaCloud allows you to remotely monitor, update and manage your device. You can add your first 10 devices for free!)
Be sure to select "Nvidia Jetson Nano" as the device type. click "add device" to download the image and burn it to an SD card using Etcher. Alternatively, you can use the balena CLI to push this code to your application.
Once the build is complete and the image has downloaded to your device, you can access OpenDataCam from your browser using the IP of your Nano. For instance: http://192.168.1.67 (it is currently set to use port 80, but you can change that with a device variable, see below.)
You can set the following device variables in the balenaCloud dashboard to modify your OpenDataCam configuration:
VIDEO_INPUT - set to usbcam for an attached USB camera (default value) or remote_cam for an IP camera. If you set to remote_cam you need to also set the variable CAM_IP. The value file is also valid, in which case you also need to set the variable VIDEO_FILE. (Note that raspberrycam is not supported currently)
CAM_IP - enter the entire IP/URL of a video stream, for instance rtsp://192.168.1.168/0- can be anything supported by OpenCV, such as .m3u8, MJPEG, etc...
CAM_DEV - sets the device name for the USB camera to use if VIDEO_INPUT is set to usbcam - default value is video0, in other words the camera device would be /dev/video0 - you should not enter the /dev/ part which is assumed.
VIDEO_FILE - set the full path and filename of a video file to use as the input when the value of VIDEO_INPUT is set to file.