| CPC G06V 20/56 (2022.01) [G06F 18/2148 (2023.01); G06N 3/082 (2013.01); G06V 10/82 (2022.01); G06V 20/58 (2022.01); G08G 1/048 (2013.01)] | 20 Claims |
|
1. A computer-implemented method of operating a neural network device including a first neural network comprising a plurality of layers, the computer-implemented method comprising:
receiving sensing data from at least one sensor;
determining, by a processor, environmental information, based on the received sensing data;
determining, by the processor, multiple layers in the plurality of layers of the first neural network, the multiple layers corresponding to the determined environmental information; and
generating a second neural network for the neural network device by dynamically changing at least two layers, among the plurality of layers of the first neural network, to the determined multiple layers,
wherein the environmental information corresponds to at least one of weather and a place, and
wherein, the environmental information corresponds to a variation of the weather and the received sensing data is an image including the variation of the weather, the computer-implemented method further comprises:
performing an image processing operation of the image and removing the variation from the image, by selecting a preprocessing layer corresponding to the environmental information among a group of preprocessing layers and by using the preprocessing layer,
generating an input feature map based on the image processing operation, the input feature map representing the weather without the variation, and
extracting a feature point from the input feature map.
|