CPC G06V 20/582 (2022.01) [G06N 3/045 (2023.01); G06N 3/063 (2013.01); G06N 3/08 (2013.01); G06N 7/00 (2013.01); G06N 20/00 (2019.01); G06V 10/454 (2022.01); G06V 10/82 (2022.01); G06V 20/584 (2022.01)] | 20 Claims |
1. A method comprising:
receiving, at a data input of a plurality of units of a connectionist network, a plurality of input picture elements, the plurality of input picture elements representing a raw image of a traffic environment including one or more traffic signs acquired by a multi-channel image sensor, the image sensor being a red-clear-clear-clear sensor;
processing, by a first unit of the plurality of units, the plurality of input picture elements with a first set of processing parameters to generate a first portion of output picture elements, the first portion of output picture elements representing a first channel of the multi-channel image sensor;
processing, by one or more additional units of the plurality of units, the plurality of input picture elements with at least a second set of processing parameters to generate one or more additional portions of output picture elements, the one or more additional portions of output picture elements representing one or more additional channels of the multi-channel image sensor;
outputting, at a data output of the plurality of units of the connectionist network, the first portion of output picture elements and the one or more additional portions of output picture elements, the first portion of output picture elements and the one or more additional portions of output picture elements being representative of the raw image; and
training, using the first portion of output picture elements and the one or more additional portions of output picture elements, another connectionist network to detect the one or more traffic signs in other raw images of other traffic environments.
|