| CPC G06V 10/82 (2022.01) [G06N 3/048 (2023.01); G06V 10/764 (2022.01); G06V 10/7715 (2022.01)] | 20 Claims |

|
1. A method for processing image information of an imaging sensor of a vehicle in an artificial neural network which includes at least one encoder and one decoder, wherein the artificial neural network solves at least one of a classification task with a plurality of classes or a regression task in which numerical output information quantized according to a plurality of quantization steps is provided, wherein the artificial neural network outputs multiple feature maps at an output interface of the artificial neural network for performing a driving assistance function, wherein allocations of image regions of the image information to classes or numerical output information quantized regarding the image information are output by the feature maps in an encoded manner such that respective inputs in corresponding matrix fields of the feature maps together produce a code word and wherein an information compression is achieved by the encoding, wherein an output number of feature maps is smaller than a number of classes or a number of quantization steps, wherein the inputs of the feature maps which form the code word are distorted by estimation errors and are converted by a decision function into a binary-coded code word, wherein the following decision function is used:
![]() wherein pi is a binary number which forms one digit of the binary-coded code word, p is the respective value of the input in the feature map, which corresponds to the digit of the binary-coded code word, and t is a decision threshold.
|