Improved safety belt detection method

Abstract

The invention provides an improved safety belt detection method. A convolutional neural network (CNN) is used as a training model and is used for solving the problem of an existing deep learning safety belt detection method that the detection accuracy is low. The detection precision of the convolutional neural network is improved through utilizing a novel feedback increment type convolutional neural network training method and a novel multi-branch final evaluation value acquisition method; meanwhile, a method for randomly selecting a safety belt target candidate region in a multi-scale manner is used, and the selecting rate of a safety belt region is increased; and finally, a method for setting a fault-tolerant threshold value by a user is utilized so that the flexibility of detection operation is improved. The improved safety belt detection method is successful application of a CNN structure to safety belt detection; and compared with an existing algorithm, the detection accuracy is improved.

Description

Technical field [0001] The invention belongs to the machine learning theory and application sub-fields of the computer application technology field, and focuses on the safety belt detection problem in the intelligent transportation technology, and is specifically an improved safety belt detection method. Background technique [0002] After an in-depth investigation of the existing seat belt detection technology, it is found that the most popular seat belt detection method is the seat belt detection algorithm based on Canny edge detection and cascaded adaboost. The entire algorithm first locates the driver's area to achieve seat belt detection. . In order to realize the driver area positioning part, the algorithm mainly converts the image to be detected into the HSV space, and then uses two linear filters in the horizontal and vertical directions to calculate the projection of the image in the horizontal and vertical directions, and comprehensively compares the projection modes to...

AI Technical Summary

Problems solved by technology

Then use the adaboost algorithm to build a classifier for seat belt detection. The problem with this method is that it is greatly affected by the noise of a single image. Image noise not only has a great impact on contour acquisition but also on license plate positioning. In addition, this method is not suitable for poor lighting. The judgment error rate of the image acquired under the condition is high, because the feature acquisition method used to construct the classifier is easily affected by external factors, the robustness of this method is not high

However, the inventive method has deficiencies in detection accuracy and time efficiency. This method uses an 8-layer CNN model, which takes a long time to train and has low algorithm efficiency.

Images