ROAD SECTION LEVEL OF SERVICE EVALUATOR SYSTEM USING CCTV CAMERA BASED ON THE DEEPSORT ALGORITHM IN OBJECT TRACKING
Transportation is an aspect that is closely related to the mobilization of both people and logistics. Of course, along with the times, technology is increasingly growing to suit human needs. The intelligent transportation system is one of the impacts of technological developments. The transportation...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/79426 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Transportation is an aspect that is closely related to the mobilization of both people and logistics. Of course, along with the times, technology is increasingly growing to suit human needs. The intelligent transportation system is one of the impacts of technological developments. The transportation system that was originally controlled in a pre-determined manner is now oriented towards artificial intelligence. One of the aspects needed in building artificial intelligence is information. In the context of transportation, the required information needs to be collected before making a decision. Existing information can inform matters that are crucial in describing current conditions such as the saturation of traffic which is certainly a factor in determining the level of service.
To collect such information, sensors are needed that can collect relevant information so that the information can be processed by the system in making decisions. In this final project, a system based on object tracking will be built to evaluate the level of service of a road section. Using the DeepSORT algorithm, a sensor was built to collect information in traffic flow and vehicle speed through traffic monitoring video sources. This sensor achieved a high and consistent detection speed with an average of 91.2 fps and a standard deviation of 4.91 fps. Apart from that, the error value for traffic flow measurements and the degree of saturation did not exceed 5% and the weighted F1-score from three experiments reached a value of 98.2%; 95.0%; and 94.2%. Speed detection performance has reached an average accuracy value of 92% with a standard deviation of 2.97 km/h for a test speed of 30 km/h. The level of service can be determined with a short sampling time of 10 minutes.
|
|---|