A vehicle front obstacle detection method based on YOLO

An obstacle detection and vehicle technology, applied in the field of obstacle detection in front of a vehicle, can solve the problems of slow detection speed and low detection accuracy, and achieve the effects of fast detection speed, high detection accuracy and simple structure

Inactive Publication Date: 2019-03-08
BEIJING INFORMATION SCI & TECH UNIV
View PDF5 Cites 25 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve the problems of low detection accuracy and slow detection

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A vehicle front obstacle detection method based on YOLO
  • A vehicle front obstacle detection method based on YOLO
  • A vehicle front obstacle detection method based on YOLO

Examples

Experimental program
Comparison scheme
Effect test

Example Embodiment

[0021] Specific implementation mode one: combine figure 1 Describe this embodiment, this embodiment is based on the specific process of the obstacle detection method in front of the vehicle based on YOLO:

[0022] Step 1. Obtain a data set. The data set is divided into a training set and a test set. Use a label box to mark each target in the training set image, and obtain the position information and category information of each target in the training set image;

[0023] Step 2. Initialize the convolutional neural network, and input the training set with the marked box into the convolutional neural network;

[0024] Step 3, preprocessing the training set images;

[0025] Step 4, dividing the preprocessed training set images into grids;

[0026] The grid divided into YOLO is responsible for the task of detecting target objects;

[0027] Step 5, the grid selects the initial candidate frame;

[0028] Randomly generate two initial candidate boxes for each grid, or define its i...

Example Embodiment

[0033] Specific embodiment 2: The difference between this embodiment and specific embodiment 1 is that the data set is obtained in the step 1, the data set is divided into a training set and a test set, and each target in the training set image is marked with a label frame to obtain the training Collect the position information and category information of each target in the image; the specific process is:

[0034] Step 11, intercept 10,000 images from the driving recorder as a data set, 8,000 images in the data set are used as a training set, and the remaining 2,000 images are used as a test set;

[0035] The images in the data set are the road conditions in front of the driver, and the images include three types of targets: pedestrians, cyclists and cars;

[0036] Step 12, use the label frame to mark each target (pedestrian, cyclist or car) in the training set image, and obtain the position information and category information of each target in the training set image;

[003...

Example Embodiment

[0040] Embodiment 3: The difference between this embodiment and Embodiment 1 or 2 is that the convolutional neural network is initialized in the step 2; the specific process is:

[0041] The YOLO detection network includes 24 convolutional layers and 2 fully connected layers (24 convolutional layers followed by 2 fully connected layers);

[0042] Among them, the convolutional layer is used to extract image features, and the fully connected layer is used to predict the image position and category probability value;

[0043] Convolutional neural network parameters include impulse Momentum, weight decay Decay, maximum number of iterations Maxbatches, learning rate Learning rate, learning rate change iterations Steps, and learning rate change ratio Scales.

[0044] Define the training parameters in the convolutional neural network as shown in Table 1;

[0045] Table 1 Network configuration parameters

[0046] Momentum

[0047] Other steps and parameters are the same as...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a vehicle front obstacle detection method based on YOLO. The objective of the invention is to solve the problems of low detection accuracy and low detection speed of an existing target detection algorithm. The method comprises the following steps of 1 acquiring a data set, and labeling each target in a training set image by using a labeling box; 2 initializing a convolutional neural network, and inputting the training set marked by the marking box into the convolutional neural network; 3 preprocessing the images of the training set; 4 dividing into 7*7 grids; 5 selecting an initial candidate box by the grid; 6 obtaining the category confidence of the target category M; 7 setting the output of the convolutional neural network according to the category confidence coefficient to obtain a final prediction box; 8 obtaining a final weight and the trained convolutional neural network; and 9 testing the images of the test set by using the trained convolutional neural network, and determining barriers in front of the vehicle. The method is applied to the field of vehicle front obstacle detection.

Description

technical field [0001] The invention relates to a method for detecting an obstacle in front of a vehicle. Background technique [0002] With the continuous development of artificial intelligence and the gradual improvement of market demand, unmanned driving has gradually become a hot research topic for experts at home and abroad. Obstacle detection in front of the vehicle is an important link in the unmanned driving system. In real traffic scenarios, target detection is affected by many factors, such as: illumination, occlusion, etc. How to quickly and accurately identify and locate the target in front of the vehicle in a complex traffic scene is related to the safety of the unmanned driving field and is a topic worthy of further study. [0003] With the continuous development of deep learning, researchers began to apply it in the field of object detection, which brought great changes to the entire field. The convolutional neural network can extract image features very we...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V20/58G06V2201/07G06N3/045G06F18/214
Inventor 王超
Owner BEIJING INFORMATION SCI & TECH UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products