CSI Camera capture, format conversions.


My intent is to read frame (from CSI camera ) into cv::Mat, for further processing.

  1. What is best way to read frame in BGR or ARGB format without any conversion ?

  2. If format conversion need to be done, then I believe NvVideoConverter is the way to go. Is there any doc that describes this class ? I could not get much info the doxygen docs of code. Specifically I am looking to learn about capture plane and output plane.

Is output plane of this class actually receives input from video source ? Does capture plane gives converted output ? This is what I tend to infer from examples like 11_camera_object_identification where capturePlane callback tends to feed to opencv consumer.


Hi Nvidia Folks,
Could one of you respond please ?

We have given information

It is the optimized way for opencv+Argus