frame > float[] > render float[]

[mmpinso]

Hi, After lots of time trying to find a camera preview approach for fast video processing, this one seems the best. Our goal is to

• capture each frame using this library
• get the rbg values into a float[]
• send it to a tensorflow graph & get it back
• render the new float[] using this library, like overriding the OriginalFilter

Could you please help us finding the fastest way to achieve the 2nd and 4th steps ?