Search

KR-102963469-B1 - AI Deep Learning-Based Smart School Zone Monitoring System and Method with Re-Identification Technology

KR102963469B1KR 102963469 B1KR102963469 B1KR 102963469B1KR-102963469-B1

Abstract

The present invention relates to an AI deep learning-based smart child protection zone monitoring system and method with re-recognition technology applied, comprising: a front and rear camera for capturing an object; an image collection unit for collecting object images from the front and rear camera; a deep learning module for analyzing collected object images to perform object detection, classification, and tracking; a preprocessing module for performing preprocessing of object images, extracting external features and license plate features, and merging the extracted information to determine whether they are the same object; a database for storing information regarding violations; and a deep learning event view module for visually providing analyzed violation information.

Inventors

  • 석동호
  • 이재민
  • 박성욱
  • 정재호

Assignees

  • 주식회사 포딕스시스템

Dates

Publication Date
20260513
Application Date
20250812

Claims (8)

  1. Front and rear cameras (110) for capturing objects; An image collection unit (120) that receives and collects object images from the above front and rear cameras (110); A deep learning module (130) that analyzes collected object images to perform object detection, classification, and tracking; A preprocessing module (140) that performs preprocessing of an object image, extracts external features and license plate features, and then merges the extracted information to determine whether they are the same object; A database (150) that stores information regarding violations; and It includes a deep learning event view module (160) that visually provides analyzed violation information, and The above front and rear cameras (110) are installed based on the road direction within the child protection zone to acquire a front image and a rear image for the direction of travel of the vehicle, respectively. The above image collection unit (120) collects object images input from the front and rear cameras (110) in frame units and includes time information in the collected image frames. The deep learning module (130) analyzes the collected object image to calculate the movement time and distance of the object, calculates the instantaneous speed and segment speed based on the calculated result, and detects the case where the object moves beyond the stop line after a delay time set based on the time when the red light of the traffic light included in the collected object image is turned on. The above preprocessing module (140) includes an image enhancement unit (141) that performs high-resolution restoration, low-light correction, and noise removal for an object image, and an object re-recognition unit (142) that extracts external features and license plate features from an object image and merges the extracted feature information to determine whether they are the same object. The object re-recognition unit (142) calculates the similarity of external features using a Siamese CNN (Siamese Convolutional Neural Network), recognizes license plate features based on optical character recognition (OCR), and then merges the external features and license plate features to determine whether they are the same object. The above database (150) stores at least one of the violation type, time of occurrence, movement path, license plate information, and speed information of the detected object as metadata, and The deep learning event view module (160) visualizes the movement path of each object based on a map or a planar layout, displays a trajectory by continuously connecting movement coordinates corresponding to the object ID, and distinguishes and expresses instantaneous speed changes, stop/acceleration events, and whether the section speed is exceeded on the movement path using multiple colors or icons. The deep learning event view module (160) plays back object images captured from the front and rear cameras (110) in a frame-by-frame manner, or outputs a captured image at the time of violation occurrence, The deep learning event view module (160) displays violation records accumulated per vehicle in a timeline format or classifies them by violation type, The deep learning event view module (160) provides, in the form of a popup or panel, a graph of the speed change of the object, the distance traveled, and a history of violations when a specific object is selected by a user. AI deep learning-based smart child protection zone monitoring system with re-recognition technology.
  2. delete
  3. delete
  4. delete
  5. delete
  6. delete
  7. delete
  8. Step of the front and rear cameras capturing an object; A step in which an image collection unit receives and collects object images captured through the front and rear cameras; A step in which a deep learning module analyzes collected object images to perform object detection, classification, and tracking; A step in which a preprocessing module performs preprocessing of an object image, extracts external features and license plate features, and merges the extracted information to determine whether they are the same object; A step of storing information regarding violations in a database; and The deep learning event view module includes the step of visually providing analyzed violation information; and The above-mentioned front and rear cameras are installed based on the road direction within the child protection zone to acquire front and rear images, respectively, regarding the direction of travel of the vehicle, and The above image acquisition unit collects object images input from the front and rear cameras on a frame-by-frame basis and includes time information in the collected image frames, and The deep learning module analyzes collected object images to calculate the movement time and distance of the object, calculates instantaneous speed and segment speed based on the calculated results, and detects cases where the object moves beyond the stop line after a set delay time has elapsed based on the time when the red light of the traffic light included in the collected object images is turned on. The above preprocessing module includes an image enhancement unit that performs high-resolution restoration, low-light correction, and noise removal for an object image, and an object re-recognition unit that extracts external features and license plate features from an object image and merges the extracted feature information to determine whether they are the same object. The object re-recognition unit calculates the similarity of external features using a Siamese CNN (Siamese Convolutional Neural Network), recognizes license plate features based on Optical Character Recognition (OCR), and then merges the external features and license plate features to determine whether they are the same object. The above database stores at least one of the violation type, time of occurrence, movement path, license plate information, and speed information of the detected object as metadata, and The above-described deep learning event view module visualizes the movement path of each object based on a map or a planar layout, displays a trajectory by continuously connecting movement coordinates corresponding to the object ID, and distinguishes and expresses instantaneous speed changes, stop/acceleration events, and whether the section speed is exceeded along the movement path using multiple colors or icons. The deep learning event view module plays back object images captured from the front and rear cameras in a frame-by-frame manner, or outputs a captured image at the time of violation occurrence, and The above deep learning event view module displays violation records accumulated on a vehicle-by-vehicle basis in a timeline format or classifies them by violation type, and The above deep learning event view module provides a graph of the object's speed change, distance traveled, and violation history in the form of a popup or panel when a specific object is selected by a user. AI deep learning-based smart child protection zone monitoring method with re-recognition technology.

Description

AI Deep Learning-Based Smart School Zone Monitoring System and Method with Re-Identification Technology The technical concept of the present disclosure relates to a smart child protection zone monitoring system and method, and more specifically, to a smart child protection zone monitoring system and method that detects and analyzes objects based on re-recognition technology and AI deep learning to determine violations. With the recent advancement of intelligent public infrastructure for enhanced traffic safety and accident prevention technologies within child protection zones, smart monitoring systems that automatically detect vehicles and pedestrians and analyze traffic violations are being introduced. Conventional smart monitoring systems have primarily relied on methods that perform license plate recognition and object detection based on object images captured by front and rear cameras, and determine whether violations have occurred. However, these conventional systems suffered from a problem where recognition rates significantly deteriorated under various outdoor conditions, such as reduced image quality, low-light environments, and damaged license plates. In particular, there was a limitation in that the reliability of violation detection was reduced because it was impossible to determine whether the objects were the same when license plate recognition failed. Therefore, there is a growing need for technology that can accurately and consistently detect violations within child protection zones in various environments by combining and merging the external features and license plate features of an object to more reliably determine whether they are the same object, and by performing integrated object detection, tracking, speed calculation, and signal violation detection based on AI deep learning. FIG. 1 is a diagram illustrating the overall configuration of an AI deep learning-based smart child protection zone monitoring system (100) to which recognition technology according to an embodiment of the present disclosure is applied. FIG. 2 is a flowchart illustrating the process of collecting object images in sequence through the front and rear cameras (110) and the image collection unit (120). FIG. 3 is a flowchart illustrating the process of object detection, classification, and tracking being performed in sequence through a deep learning module (130). FIG. 4 is a flowchart illustrating the process of merging external shape and license plate feature extraction information and determining whether they are the same object in sequence through the preprocessing module (140). FIG. 5 is a flowchart illustrating the process of visually providing violation information through the deep learning event view module (160) in sequence. FIG. 6 is a diagram illustrating the components of a control system (200) that controls an AI deep learning-based smart child protection zone monitoring system (100) to which re-recognition technology is applied according to one embodiment. FIG. 7 is a flowchart illustrating an AI deep learning-based smart child protection zone monitoring method with recognition technology applied according to an embodiment of the present disclosure in a series of sequences. Hereinafter, various embodiments of the present disclosure are described in conjunction with the accompanying drawings. As various embodiments of the present disclosure may be subject to various modifications and may have various forms, specific embodiments are illustrated in the drawings and described in detail. However, this is not intended to limit the various embodiments of the present disclosure to specific forms, and it should be understood that they include all modifications and/or equivalents and substitutions that fall within the spirit and scope of the various embodiments of the present disclosure. In relation to the description of the drawings, similar reference numerals have been used for similar components. In various embodiments of the present disclosure, terms such as “comprising” or “having” are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof. In various embodiments of the present disclosure, expressions such as “or” include any and all combinations of the words listed together. For example, “A or B” may include A, may include B, or may include both A and B. Expressions such as "first," "second," "first," or "second" used in various embodiments of the present disclosure may modify various components of the various embodiments, but do not limit such components. For example, such expressions do not limit the order and/or importance of such components and may be used to distinguish one component from another. When it is mentioned that a component is "connec