Abstract
Traditional lane detection algorithms use lane features and mathematical models to detect the location of lane lines, which is sensitive to complex scenes and multiple interference conditions. Full Convolutional Neural Network (FCN) considers the full-connected layer in the traditional network as convolution layer, and connects the deconvolution layer. After upsampling, we can obtain the segmentation graph of the pixels. The feature extraction network is based on AlexNet, which reduces the receptive field by reducing the convolution kernel size and downsampling ratio of the first layer and introduces the conditional random field (CRF) to enhance the image edge constraints. Finally, the CRF is incorporated into the FCN, and we construct an end-to-end network that is categorized by pixel-by-pixel prediction. 367 road scenes in CamVid dataset were used in this experiment, and 11 types of pixels are classified. The accuracy rate of training is 90.1%, and the test result is good.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.