Introduction
Gachon University operates an eco-bus (hereinafter referred to as a moodang) to help move between buildings located in the mountains. Recently, the number of moodang has increased explosively with the creation of a new building AI hall on the middle of the campus mountain. Due to the increase in the number of users, a service that allows moodang to check the location of moodang and the congestion of the stops in advance has begun to be needed.
We implemented this project to address these requirements.
This project is
- Only use cameras that are cheaper than GPS for the price.
- Extract the location information of the moodang on campus using only the camera.
- Detect people waiting at the stop and extract congestion and number of people.
- The provided information is provided to users through the web page in real time.
The Location Detection and Congestion Detection project is designed to allow moodang to calculate the congestion level of each bus's current location and the final stop and check it through a web page. So when you make a project and experiment, the experiment location is Gachon University Compass. In this project experiment, we used a device with two cameras and a smartphone to run a web page. The camera is considered to have a moodang installed in front of the bus and across from the main stop. Web pages are provided to users by moodangs through hosted web pages. Devices to which the camera is connected are connected to the real time database of the firebase.
The moodang's front camera (Camera type 1) captures the situation in front of the moodag. Camera type 1 detects objects located on the route of the moodang. The detected objects are identified by tracking. The identified objects are recorded for a fixed time and stored temporarily, and the current position is predicted based on the identified objects at a fixed time. The predicted location sends the id value of the installed moodang and the predicted location value to the server.
A camera (Camera type 2) installed across the main stop tracks people. Camera type 2 detects everyone seen at the station. Tracking is used to identify people waiting at the station among those detected. Assign an id value to the same person, and if a person with that id value is detected for more than a certain period of time within the interval taken by Camera type 2, the person with that id value is classified as a waiting person. Extract congestion information by dividing the number of people classified as waiting by congestion level criteria. Sends the extracted congestion level information and the number of people waiting to the server with a value called the main gate stop.
The web page screen provides users with one information button and two slides. Click the Information button to provide information, driving information, and congestion classification criteria through text and images. Without clicking the Information button, the first slide you see provides information on the route from the main gate to the AI hall (upward) and the second slide provides information on the route from the AI hall to the main gate (downward). Displays the location, congestion, and number of people on standby in real time for the main stop and midpoint positions, respectively, up and down.