I am currently learning some stereo vision applications (i.e. depth mapping, disparity mapping …). I have written a desktop application using C++ and OpenCV in order to process stereo images and video flux.
The goal of the project is to use an embedded system which will at least process depth maps in “real time”. Firstly, two cameras will have to send the video flux to my desktop application which will post process it. Secondly, image processing will have to be implemented on the board.
I am not sure about the resolution of the captured images for now, but I think I will go with 320x240 or 640x480. Ideally, I would like both cameras to operate at 20-30 FPS.
In order to carry out my project successfully, I need to choose a capable enough embedded device which will do the job.
I am considering using 2 CSI cameras in order to achieve stereo vision
I currently do not own a kit and I would like to know if the NVIDIA Jetson TK1 is a reliable board for this kind of application.
Thanks for your answers