Automatic driving scene-oriented semantic segmentation method based on deep learning
A technology for automatic driving and semantic segmentation, applied in the field of deep learning, can solve problems such as the difficulty of segmentation of small target boundaries, and achieve the effect of reducing computational complexity, better effect, and fast detection speed.
Inactive Publication Date: 2021-03-16
TIANJIN UNIV OF SCI & TECH
View PDF5 Cites 10 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] This paper proposes a semantic segmentation method based on the encoder-decoder model in the autonomous driving scene, and uses deep learning technology to solve the difficult problem of small target boundary segmentation
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0024] The present invention is described below in conjunction with accompanying drawing.
[0025] Flowchart such as figure 1 The research on the semantic segmentation method based on deep learning in the automatic driving scene shown mainly includes the following steps.
[0026] Step 1: Get Cityscapes (5000 finely labeled images) and Camvid two datasets as training set and test set. Label the data set and convert it into tfrecord format data that can be easily obtained by Tensorflow.
[0027] Step 2: Divide the data set into a training set and a test set with a ratio of 7:3. In the Tensorflow environment based on the Python programming language, geometric transformations such as flipping, rotating, scaling, and shifting are performed on the images of the two datasets to enhance image data, and CUDA is used for accelerated operations.
[0028] Step 3: Use the improved Xception to pre-train the weights on Cityscapes and Camvid to complete the extraction process of the shape,...
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
Login to View More Abstract
The invention discloses an automatic driving scene-oriented semantic segmentation method based on deep learning, and mainly solves the problems of large calculation amount and low segmentation accuracy in a current streetscape image semantic segmentation technology. According to the method, Cityscapes and Camvid data sets are used as a training set and a test set, the data sets are preprocessed ina Tensorflow environment, then an improved Xception classification model is used as a trunk network, feature extraction is performed on a target object in a complex scene image, an Xception recognition processing result is sent to DeeplabV3 + for semantic segmentation, and finally, a segmentation result is obtained after training, testing and network parameter adjustment. In the method, the improved Xception is used as a classification network model, so that the accuracy of image target recognition and segmentation is improved, and the recognition time and economic cost are reduced. The method can be applied to the fields of automatic driving, military affairs and the like.
Description
technical field [0001] The invention relates to the fields of deep learning, computer vision, and image analysis, and specifically relates to a semantic segmentation method based on deep learning for automatic driving scenes. Background technique [0002] In recent years, with the rapid development of artificial intelligence technology, people's lives are becoming more and more intelligent, more and more intelligent products have come out one after another, and people's lives are increasingly relying on intelligent products to complete some trivial tasks. Nowadays, a large number of intelligent equipment and technologies such as assisted driving cars, drones, robots, and city virtualization have been developed, and the demand for intelligent recognition has become more and more urgent. However, image segmentation is the basis and premise of intelligent recognition, and the effect of segmentation directly affects the efficiency and accuracy of recognition. At the same time, ...
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
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06T7/12G06K9/62G06N3/04G06N3/08
CPCG06T7/12G06N3/08G06T2207/20081G06T2207/20084G06T2207/20192G06N3/045G06F18/214G06F18/241
Inventor 赵继民许俊辉王颖林丽媛腾万伟向炼郝迪韦赛远
Owner TIANJIN UNIV OF SCI & TECH