Abstract
Increased computing power and advanced deep learning technology have enabled computers to effectively deal with problems that cannot be solved by ordinary people. Many attempts have been made to utilize deep learning technology to analyze road images and efficiently control crossroad vehicle flow.
In this research, a new methodology is proposed for identifying the number of vehicles on the road using CNN (convolution neural network), deep learning technology that specializes in image classification. Unlike previous studies that used regression methods and video frames as input, this study determined the number of vehicles using real-time photographic images and classification methods for one lane. An experiment was conducted to find the optimal combination of variables using sensitivity analysis. The optimal network determined the number of vehicles on one lane with a high accuracy of 98.31%.
In this research, a new methodology is proposed for identifying the number of vehicles on the road using CNN (convolution neural network), deep learning technology that specializes in image classification. Unlike previous studies that used regression methods and video frames as input, this study determined the number of vehicles using real-time photographic images and classification methods for one lane. An experiment was conducted to find the optimal combination of variables using sensitivity analysis. The optimal network determined the number of vehicles on one lane with a high accuracy of 98.31%.
| Translated title of the contribution | Counting Algorithm Structure for Waiting Vehicles by using CNN |
|---|---|
| Original language | Korean |
| Pages (from-to) | 176-181 |
| Number of pages | 6 |
| Journal | 한국생산제조학회지 |
| Volume | 29 |
| Issue number | 3 |
| DOIs | |
| State | Published - Jun 2020 |