I am designing a professional imaging system using TX1 and or TK1 that will be a professional industrial product. But I am a newbie and at the very early design stage.
I am planning a system that reads camera inputs, process (detection, tracking), than encodes to store or transmit.
There are many usb 3.0 cameras available (ptgrey, lumenera, ximea, basler etc.) that seems to support jetson tx1/tk1 platforms (am i wrong?). Some of these cameras can give RGB data, so that it seems no de-bayering needed (Should I do it myself anyway?).
Would it be any better to use CSI-2 lanes since they include ISPs? Why? It seems that CSI-2 is for mobile cameras not for larger camera systems. Should I (somehow) use it anyway?
Or should I even not design a system that needs professional camera processing with jetsons? I believe there are enough power room for computer vision processing and encoding at the same time but I cannot be sure about the support for camera drivers etc.
I would really appreciate help with the camera system. Can somebody point me in right direction? I am a bit confused.
Thanks in advance.