Driver behavior secondary monitoring method based on model quantification

Abstract

The invention provides a driver behavior secondary monitoring method based on model quantification. The method comprises the following steps: acquiring offline video data, wherein the off-line video data is a part where a driver behavior event occurs in the on-line video data; running a first video understanding network, and taking the offline video data as the input of the first video understanding network, wherein the first video understanding network is quantized by a second video understanding network; and obtaining a secondary monitoring result of the driver behavior according to the event monitoring task score output by the first video understanding network. According to the invention, the problem of high false alarm rate of driver behavior event monitoring caused by insufficient computing resources of the vehicle equipment end is effectively solved; and meanwhile, the offline video feature extraction efficiency and the target prediction accuracy can be considered, so that the robustness is stronger when the offline video data uploaded by a plurality of vehicle-mounted equipment ends are secondarily monitored.

Description

technical field [0001] The invention relates to the technical field of deep learning, in particular to a method for secondary monitoring of driver behavior based on model quantification. Background technique [0002] Driver behavior recognition in the cockpit is the key to ensuring driver safety. The algorithm needs to monitor and identify driver behavior in real time, and issue timely alarms to reduce unsafe driving behaviors of drivers, such as smoking, making phone calls, and closing eyes due to fatigue. , yawning, camera occlusion, distraction, direct sunlight, whether the camera angle is correct, whether the driver is inside the camera, etc. [0003] In order to achieve the purpose of monitoring the driver's behavior in the cabin, the existing technology provides some algorithms that use face key points, smoking detection / classification or mobile phone detection / classification, combined with post-processing logic for behavior judgment and recognition. [0004] However,...

AI Technical Summary

Problems solved by technology

[0006] The invention provides a model-based quantification-based secondary monitoring method for driver behavior, which is used to solve the defects in the prior art that are limited by the limited computing resources of the mobile terminal equipment of the vehicle, the accuracy of the monitoring and identification results is low, and the false positive rate is high , to achieve accurate driver behavior monitoring

Images