Multi-task attribute identification method and device for driving scene, medium and equipment

Abstract

The embodiment of the invention discloses a multi-task attribute identification method and device for a driving scene, a medium and equipment. The method comprises: acquiring image data of a driver ina driving scene; inputting the image data into a pre-trained multi-task network model for attribute identification, wherein the attribute identification task of the multi-task network model comprisesmask identification, glasses identification, smoking identification and calling identification; and determining a multi-task attribute identification result of the driver according to the attribute identification output result of the multi-task network model. By implementing the technical scheme, a multi-task network model can be adopted to recognize various characteristics of a driver at the same time, and an output result capable of integrally reflecting the driving state of the driver is output, so that the purpose of improving the recognition accuracy of the state of the driver is achieved.

Description

technical field [0001] The embodiments of the present application relate to the field of computer technology, in particular to image recognition technology, and in particular to a multi-task attribute recognition method, device, medium and equipment for driving scenes. Background technique [0002] In the ADAS system, the accurate and timely provision of driver characteristic attributes is the key factor to realize the intelligentization of safe driving system. A common approach is to use different method models to identify different driver attributes separately. [0003] For the driver attribute identification method, a method based on deep learning is generally used to identify a certain attribute feature of the driver. The traditional method is to combine multiple models to identify each attribute feature of the driver. However, the traditional method has a single scene and is difficult to deal with occlusion, Scenes in the cab with poor lighting, many people, missing im...

AI Technical Summary

Problems solved by technology

[0003] For the driver attribute recognition method, a method based on deep learning is generally used to identify a certain attribute feature of the driver. The traditional method is to combine multiple models to identify each attribute feature of the driver. However, the traditional method has a single scene and is difficult to deal with Scenes in the cab with poor lighting, many people, missing important attributes, etc.

Secondly, if multiple models are used to identify driver attributes, the response speed of the system will also slow down, the size of the model will increase, and the hardware requirements will increase accordingly. In addition, in the case of multiple people, it is impossible to ensure that all attribute features come from the same person.

Finally, for the identification of smoking and phone calls, most of the current methods use the anchor-based neural network model, which not only cannot be integrated with other attributes into a model, but also increases the difficulty of calculation and labeling

Images