Abstract: An apparatus includes a preprocessing network and a quantized neural network. The preprocessing network may be configured to (a) receive an input, (b) perform a transform operation on the input, and (c) generate an intermediate result in response to the transform operation. The quantized neural network may be configured to (a) receive the intermediate result, (b) perform a plurality of weighted operations on the intermediate result, and (c) generate an output in response to the weighted operations. The quantized neural network is generally adapted to a hardware constraint of an edge device. The weighted operations generally have a quantization loss corresponding to the hardware constraint. The transform operation performed by the preprocessing network is generally configured to provide compensation for the quantization loss.

Abstract: A 3D display is characterized by a quality of viewing experience (QVE) mapping which represents a display-specific input-output relationship between input depth values and output QVE values. Examples of QVE mappings based on a metric of “viewing blur” are presented. Given reference depth data generated for a reference display and a representation of an artist's mapping function, which represents an input-output relationship between original input depth data and QVE data generated using a QVE mapping for a reference display, a decoder may reconstruct the reference depth data and apply an inverse QVE mapping for a target display to generate output depth data optimized for the target display.

Abstract: A 3D display is characterized by a quality of viewing experience (QVE) mapping which represents a display-specific input-output relationship between input depth values and output QVE values. Examples of QVE mappings based on a metric of “viewing blur” are presented. Given reference depth data generated for a reference display and a representation of an artist's mapping function, which represents an input-output relationship between original input depth data and QVE data generated using a QVE mapping for a reference display, a decoder may reconstruct the reference depth data and apply an inverse QVE mapping for a target display to generate output depth data optimized for the target display.