A real-time vehicle logo detection method based on multi-scale feature fusion and DCNN

Abstract

The invention discloses a real-time vehicle logo detection method based on multi-scale feature fusion and DCNN. The method comprises the following steps of collecting and screening pictures; making adata set, making a vehicle logo data set according to a deep learning standard VOC data set format; designing a network, taking a YOLO framework as a basis, taking an improved Darknet-20 network as abasic network, carrying out the channel fusion on feature maps with different depths, and building a network model; carrying out model training, training the vehicle logo data set by utilizing a network model, and carrying out parameter setting, data enhancement and multi-scale training during model training; and testing and evaluating the model. According to the invention, an end-to-end one-stagenon-cascade structure is used to process the vehicle logo detection as a regression problem, so that the improved network structure can better adapt to the detection of large and small vehicle logosand similar vehicle logos under various scenes, especially has very good robustness for the detection of small vehicle logo targets, and greatly improves the vehicle logo detection speed, recall ratioand accuracy.

Description

technical field [0001] The invention relates to the technical field of target detection in the direction of computer vision, in particular to a real-time vehicle logo detection method based on multi-scale feature fusion and DCNN. Background technique [0002] With the continuous development of the economies of countries in the world, the types and quantities of private cars are also increasing. As a common means of passenger and transportation, automobiles provide convenience to people, and their effective supervision in roads, communities, parking lots, etc. has become an urgent problem to be solved. With the trend of world globalization and informatization development, manual supervision is gradually replaced by intelligent transportation systems. Through advanced image acquisition, processing and intelligent analysis technologies, vehicle detection (i.e. vehicle positioning and identification) and its attribute identification have become more and more efficient and accur...

AI Technical Summary

Problems solved by technology

[0009] (1) Most of the car logos are small objects, which are more difficult in feature extraction

[0010] (2) Car logo detection is easily affected by factors such as car logo image resolution, rotation angle, saturation, exposure, and hue

[0011] (3) The size and proportion of the car logo in different photos are different, and the convolutional neural network has a large difference in the proportion of target detection, and the generalization ability of the target detection is poor.

[0012] (4) Previous car logo detection algorithms have preprocessed the car logo pictures, ignoring the autonomous learning ability of the neural network on the original car logo pictures

[0013] (5) The above four reasons lead to high complexity, low accuracy, low recall rate, and poor real-time performance of vehicle logo detection

Images