Urban digital twin platform system and moving object information analysis and management method therefor
Provided are an urban digital twin platform system and a moving object information analysis and management method therefor. Through the urban digital twin platform system and the moving object information analysis and management method, a moving object such as a vehicle or a pedestrian may be detected from multimodal sensor data, data on the moving object may be generated, and a situation may be quickly determined by deriving complex actions of the moving object. The urban digital twin platform system includes a multimodal sensor data input and objectification module configured to detect a moving object and to generate objectification data, a multimodal sensor data analysis module configured to classify basic actions of the moving object, to classify complex actions of the moving object, and to generate moving object information, and an urban space data server configured to store the objectification data and the moving object information.
1. An urban digital twin platform system comprising:
a multimodal sensor data input and objectification module configured to detect a moving object from data collected by a sensor and to generate objectification data that is data on the moving object;
a multimodal sensor data analysis module configured to classify basic actions of the moving object on a basis of the objectification data, to classify complex actions of the moving object through pattern analysis on the basic actions, and to generate moving object information; and
an urban space data server configured to store the objectification data and the moving object information,
wherein, when the data collected by the sensor is image data, the multimodal sensor data input and objectification module acquires a position vector of the moving object by converting pixel coordinates where the moving object is located in the image data into (longitude, latitude, and altitude).
2. The urban digital twin platform system of claim 1 , further comprising:
a CCTV recorder interface module configured to provide the multimodal sensor data input and objectification module with an interface for accessing image data stored in a CCTV recorder.
3. The urban digital twin platform system of claim 1 , further comprising:
an urban space data visualization module configured to extract the objectification data and the moving object information, which are stored in the urban space data server, according to time and spatial conditions, and to visualize the extracted objectification data and the moving object information.
4. The urban digital twin platform system of claim 3 , wherein the urban space data visualization module is able to reproduce dynamic data including the moving object by changing viewpoints.
5. The urban digital twin platform system of claim 1 , wherein, when the data collected by the sensor is image data, the multimodal sensor data input and objectification module detects the moving object for each frame of the image data.
6. The urban digital twin platform system of claim 1 , wherein, when the data collected by the sensor is image data, the multimodal sensor data input and objectification module detects the moving object on a basis of a position and a size of the moving object displayed in the image data by using a deep learning-based multi-object tracking technology.
7. The urban digital twin platform system of claim 1 , wherein the multimodal sensor data analysis module classifies the basic actions of the moving object by using the objectification data of the moving object and information collected from outside.
8. The urban digital twin platform system of claim 1 , wherein the multimodal sensor data analysis module classifies the complex actions of the moving object by determining whether a pattern of a plurality of successive basic actions of the moving object matches a specific pattern, and the complex actions include combined basic actions.
9. The urban digital twin platform system of claim 1 , wherein the urban space data server stores the objectification data and the moving object information for each of a plurality of tiles on a basis of an installation position of a sensor having detected the moving object.
10. The urban digital twin platform system of claim 1 , wherein the urban space data server stores only the objectification data and the moving object information without storing image data collected by a camera sensor.
11. A moving object information analysis and management method comprising:
a data objectification step of detecting a moving object from sensor data and generating objectification data that is data on the moving object;
a situation analysis step of classifying basic actions of the moving object at a specific point in time on a basis of the objectification data, classifying complex actions of the moving object through situation analysis based on the basic actions of the moving object, and generating moving object information; and
a data storage step of storing the objectification data and the moving object information,
wherein, when the sensor data is image data, a position vector of the moving object is acquired by converting pixel coordinates where the moving object is located in the image data into (longitude, latitude, and altitude).
12. The moving object information analysis and management method of claim 11 , wherein the data objectification step comprises:
a sensor data reception step of receiving data collected by a sensor;
a moving object detection step of detecting a moving object from the data collected by the sensor;
a step of classifying the moving object;
a step of acquiring position data of the moving object on a basis of the data collected by the sensor; and
a step of acquiring direction data and speed data of the moving object by using the position data.
13. The moving object information analysis and management method of claim 11 , wherein the situation analysis step comprises:
a step of receiving the objectification data of the moving object;
an external information collection step of collecting information related to the moving object from outside;
a basic action classification step of classifying the basic actions of the moving object at a specific point in time by using the objectification data of the moving object and the information collected from the outside; and
a complex action classification step of classifying the complex actions of the moving object through situation analysis based on the basic actions of the moving object.
14. The moving object information analysis and management method of claim 11 , wherein the data storage step comprises:
an objectification data and object information reception step of receiving the objectification data and the moving object information;
a step of storing the objectification data and the moving object information for each of a plurality of tiles on a basis of an installation position of a sensor having detected the moving object; and
a step of updating metadata of a specific tile when a sensor is additionally installed or removed in the specific tile.