In view of the problems of many parameters, complex network and too much memory occupied by the current target detection algorithm based on convolutional neural network in the edge computing equipment, a video based lightweight convolutional neural network YOLOv5-MN3 for Abandoned Objects in Freeway is proposed. Firstly, we reduce the network parameters and the amount of computation by changing the backbone network architecture and replacing the standard convolution with the deep separable convolution. Secondly, the paper enhances the feature extraction ability of neural network by integrating attention mechanism and improves the recall and accuracy of model detection. Finally, through knowledge distillation to further compress the model, we completed the design of lightweight network. The experimental results showed that the average accuracy of YOLOv5-MN3 network can reach 88.2%, which is 38.68% smaller than the original network. Therefore, the network met the requirements of edge computing device deployment.
Roadside Light Detection and Ranging (LiDAR) can provide over the horizon perception information for connected vehicles (CV). However, its performance may be affected by the weather, especially in rainy and snowfall weather. To improve all-weather working ability, a combined denoising algorithm is proposed in this paper after analyzing the shortcomings of the existing filters. The filter is composed of crop box filter, ray ground filter, voxel filter, and statistical outlier filter. By combining multiple point clouds filters, the snowfall points are removed and the effectiveness of general filters in complex scenes is verified. The experiment shows that it not only can retain the traffic objects’ features, but also realize denoising on real point clouds data of snowfall weather.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.