Traffic sign board detection method based on improved YOLOF model

Abstract

The invention discloses a traffic sign board detection method based on an improved YOLOF model. The method comprises the steps of carrying out the augmentation of a collected traffic sign board sample, and inputting a data set after the augmentation into an improved YOLOF network model for training; and detecting the trained improved YOLOF network model, and ending the detection if the detection result is qualified. The sample set is augmented through multiple augmentation modes, a large number of samples are obtained while the collection cost is reduced, the robustness of the model and the detection performance of the traffic sign board in an unmanned complex scene are improved, the traffic sign board is detected by using the improved YOLOF model, and the detection accuracy of the traffic sign board is improved, so that the problems of missing detection and false detection of the traffic sign board in a complex scene are solved, and the detection speed of the traffic sign board in an unmanned environment is improved.

Description

technical field [0001] The invention relates to the technical field of automatic driving control, in particular to a traffic sign detection method based on an improved YOLOF model. Background technique [0002] Object detection is one of the most important tasks in the field of computer vision, usually applied in the field of autonomous driving. As a direction of future technological development, autonomous driving has become a research hotspot in recent years. Traffic sign detection is an important part of the perception module in the field of autonomous driving. It can automatically identify and mark traffic signs, and transmit the results to the autonomous driving decision-making module to ensure that vehicles can drive safely in accordance with traffic rules. [0003] Before the advent of deep neural networks, traffic sign detection usually employed methods based on feature extraction, such as features such as color and shape. Scale-invariant feature transform (SIFT) a...

AI Technical Summary

Problems solved by technology

[0005] However, traffic sign detection in unmanned driving environments in complex scenes will be disturbed by factors such as lighting changes, bad weather, and other graphics similar to traffic signs. The above traffic sign algorithms will all have the problem of missed detection and false detection. It is urgent to provide a method sufficient to solve the above problems

Images