Detection of Parking Spaces in Open Environments with Low Light and Severe Weather

DOI

10.2991/978-2-494069-31-2_354How to use a DOI?

Keywords

deep Learning; object detection; parking space detection; knowledge distillation

Abstract

Based on the goal of improving the accuracy of all-day outdoor parking lot detection, and considering the difficulty of detecting small spatial targets in many images and the problem that the detection performance at night has a large gap with that at daytime, the all-day outdoor parking lot detection algorithm is improved on the basis of the existing SSD algorithm. Firstly, the input image is amplified and sampled and data processed; then the VGG16 backbone feature extraction network of the SSD model is replaced by the ResNet101 residual network. The gradient problem that occurs with the deepening of the network training is avoided, thus enabling the feature map to extract richer image information. For the nighttime detection problem, the author is inspired by the SID model to train a new nighttime model using YOLO in one dataset, and then distill the features of the SSD model that has been trained using daytime images by SID. The new model has the potential features of SSD and YOLOV3 respectively, and can directly test outdoor parking images throughout the day, so that the new model finally has both daytime and nighttime features. The new model also has higher detection accuracy than other models by experimental comparison. Finally, the bonding layer responsible for fusing the models, reduces the total amount of computational resources, so the weights of the models are also improved.

Copyright

© 2022 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Borui Li
PY  - 2022
DA  - 2022/12/29
TI  - Detection of Parking Spaces in Open Environments with Low Light and Severe Weather
BT  - Proceedings of the 2022 6th International Seminar on Education, Management and Social Sciences (ISEMSS 2022)
PB  - Atlantis Press
SP  - 3011
EP  - 3020
SN  - 2352-5398
UR  - https://doi.org/10.2991/978-2-494069-31-2_354
DO  - 10.2991/978-2-494069-31-2_354
ID  - Li2022
ER  -