A method for analyzing a road surface condition according to the present invention includes the steps of: generating, by a learning part, a first recognition model through learning; and when the first recognition model receives a video signal for learning and an audio signal of interest for learning, and calculates an original prediction value, which is a predicted probability for each of a plurality of conditions of a road surface, generating, by the learning part, a second recognition model by processing the original prediction value into a soft label for the soft prediction value of the second recognition model.
본 발명의 노면의 상태를 분석하기 위한 방법은 학습부가 학습을 통해 제1 인식모델을 생성하는 단계와, 상기 학습부가 제1 인식모델이 학습용 영상 신호 및 학습용 관심 오디오 신호를 입력 받고, 노면의 복수의 상태 각각에 대해 예측한 확률인 원본예측값을 산출하면, 상기 원본예측값을 제2 인식모델의 소프트예측값에 대한 소프트레이블로 가공하여 제2 인식모델을 생성하는 단계를 포함한다.
Apparatus for analyzing road surface condition based on driving noise and method therefor
주행 소음 기반의 노면의 상태를 분석하기 위한 장치 및 이를 위한 방법
2023-04-04
Patent
Electronic Resource
Korean
Apparatus for analyzing road surface condition based on driving noise in tunnel and method therefor
European Patent Office | 2023
|Apparatus for analyzing road surface condition using duration of driving noise and method therefor
European Patent Office | 2023
|European Patent Office | 2023
|Apparatus for analyzing road surface condition using ultrasound and method therefor
European Patent Office | 2023
|Apparatus for analyzing road surface condition using resonance tube and method therefor
European Patent Office | 2023
|