Evaluation method for ADAS (Advanced Driver Assistant System) based on machine learning

Abstract

The invention provides an evaluation method for an ADAS (Advanced Driver Assistant System) based on machine learning. The evaluation method comprises the steps of acquiring an image sample, and building a standard space of ADAS test indexes, wherein the standard space comprises a standard fuzzy space, a standard illumination space and a standard shading space; performing machine learning on the image sample of the standard space at an off-line state to acquire an evaluation criterion of the ADAS; and acquiring an image to be tested, and evaluating the ADAS according to the acquired evaluation criterion. According to the evaluation method, evaluation can be performed on the ADAS under dynamic factors, and the accuracy of an evaluation result is effectively improved. In addition, the evaluation method is high in adaptability and can perform accurate evaluation under the conditions of mass data, multiple parameters and time varying.

Description

technical field [0001] The invention relates to an evaluation method, in particular to an evaluation method for an ADAS system based on machine learning. Background technique [0002] ADAS system is the English abbreviation of Advanced Driver Assistance System, which is the abbreviation of Advanced Driver Assistant System. Make the driver aware of possible dangers and improve the safety and comfort of driving. [0003] The ADAS system needs to be evaluated when it is used. In the prior art, the evaluation of the ADAS system is realized by the following method: firstly, the evaluation parameters are set artificially, and then the image to be tested is extracted through the image to be tested obtained by the ADAS system. Features The parameters are compared with the set evaluation parameters to obtain the corresponding performance indicators of the ADAS system. This method has the following defects: since the set parameter indicators are a standard set by humans, once set, ba...

AI Technical Summary

Problems solved by technology

[0003] The ADAS system needs to be evaluated when it is used. In the prior art, the evaluation of the ADAS system is realized by the following method: firstly, the evaluation parameters are set artificially, and then the image to be tested is extracted through the image to be tested obtained by the ADAS system. Features The parameters are compared with the set evaluation parameters to obtain the corresponding performance indicators of the ADAS system. This method has the following defects: since the set parameter indicators are a standard set by humans, once set, basically follow this setting However, the environmental factors affecting the performance of the ADAS system will also change accordingly when the vehicle is driving. Therefore, the final evaluation results of the ADAS system will be inaccurate, which will affect the final use; and the existing technology The method only has relatively referential evaluation results when the processing data is small. When dealing with massive data, rapid changes in data parameters, and complex and diverse conditions, the existing methods are obviously unable to meet the evaluation requirements.

Images