I wonder if you develop an ML model that uses brightness and color reduction cues in the video in order to estimate depth in lieu of a dive computer. I suspect cameras auto-exposure or underwater modes may nullify this.
But yea, one big issue is that it would limit how you shoot the video as well. If the camera accidentally catches the sun or a bright surface reflection, the exposure shift could throw off the depth estimation. Maybe locking exposure/white balance settings could help, but it would still be tricky. Definitely an interesting problem to tackle! Thanks for sharing! :)
noppanut15|1 year ago
But yea, one big issue is that it would limit how you shoot the video as well. If the camera accidentally catches the sun or a bright surface reflection, the exposure shift could throw off the depth estimation. Maybe locking exposure/white balance settings could help, but it would still be tricky. Definitely an interesting problem to tackle! Thanks for sharing! :)