Depth from Subpixel disparity

nickjrz

I am collecting disparity frames using the gen2-collecting-training-data and using subpixel modeand median filter. Any idea how I can calculate depth from this? The results I am getting for depth do not correlate to the data being collected

This is how I am collecting the disparity frames:

"disparity": lambda item: cv2.applyColorMap((item.getFrame() / 11.921568627 ).astype(np.uint8), cv2.COLORMAP_JET)

This is how I am calculating the depth from the disparity frames

disparity_map = disparity_dai focal = disparity_map.shape[1] / (2. * math.tan(math.radians(fov / 2))) depthScaleFactor = baseline * focal depthFrame_dai = (depthScaleFactor / disparity_map).astype(np.uint8) depth_map = disparity_map

erik

For anyone else wondering, it was answered here:

        disp_levels = 8 //(8 subpixel bits)
        with np.errstate(divide='ignore'):
            depth = (disp_levels * depthScaleFactor / disparityFrame).astype(np.uint16)