KR-20260063578-A - Pedestrian Detection Device Based on Deep Learning Algorithm

Abstract

The present invention relates to a pedestrian detection device based on a deep learning algorithm, comprising: a detection unit that captures a pedestrian on a crosswalk to acquire image information; a collection unit that collects the captured pedestrian image information; a preprocessing unit that removes noise and normalizes the collected pedestrian image information using a digital image filter; an inference unit that receives the preprocessed image information and detects pedestrian objects in real time through a YOLO8m model; and an output unit that outputs a warning sound and an LED traffic light through a control unit when the pedestrian is detected.

Inventors

• 최원혁
• 정우진
• 한조원
• 임영근

Assignees

• 한서대학교 산학협력단

Dates

Publication Date

20260507

Application Date

20241030

Claims (3)

1. A detection unit that captures pedestrians on a crosswalk to acquire video information; A collection unit for collecting the above-mentioned pedestrian video information; A preprocessing unit that removes noise and normalizes the above-mentioned pedestrian image information using a digital image filter; An inference unit that receives the above-mentioned preprocessed image information and detects pedestrian objects in real time through a YOLO8m model; and A deep learning algorithm-based pedestrian detection device characterized by including: an output unit that outputs a warning sound and an LED traffic light through a control unit when the pedestrian is detected.
2. In paragraph 1, A deep learning algorithm-based pedestrian detection device characterized by further including a control unit that executes overall control functions of the pedestrian detection device using programs and data stored in a storage unit.
3. In paragraph 1, A deep learning algorithm-based pedestrian detection device characterized by the above inference unit verifying whether an object existing within image information is a person after having learned in advance pedestrians included in the object through the YOLO8m model.

Description

Pedestrian Detection Device Based on Deep Learning Algorithm The present invention relates to a pedestrian detection device based on a deep learning algorithm, and more specifically, to a pedestrian detection device based on a deep learning algorithm that can reduce the risk of pedestrian accidents by utilizing IoT sensors and real-time signal control technology. The number of traffic accident fatalities per 100,000 children in Korea is 1.4, which is higher than the OECD average of 1.0, and the number of pedestrian fatalities per 100,000 children is 0.88, which is also higher than the OECD average of 0.31. The number of traffic accidents involving children in Korea has also been increasing over the past three years, and following the recent passage of the Min-sik Act in the National Assembly, safety measures are being strengthened to prevent accidents in child protection zones and crosswalks. Recent smart crosswalks are representative applications of pedestrian safety platforms that utilize IoT sensors and real-time signal control technology to reduce the risk of pedestrian accidents; this platform can support the construction and operation of smart crosswalks by providing a technical and policy foundation. However, in these pedestrian safety platforms, IoT sensors collect data in real time and transmit it to cloud servers; however, delays may occur depending on network speed and stability, which could hinder smooth real-time signal control. Additionally, the detection capabilities of sensors such as cameras can be affected by changes in weather or lighting, making accurate detection difficult. Furthermore, when multiple sensors operate simultaneously, they may misidentify objects other than pedestrians, such as vehicles or animals, negatively impacting signal control. FIG. 1 is a block diagram showing a pedestrian detection device based on a deep learning algorithm according to the present invention. Embodiments of the present invention are described below with reference to the attached drawings so that those skilled in the art can easily implement the invention. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein. Then, a preferred embodiment of a pedestrian detection device based on a deep learning algorithm according to the present invention will be described in detail. FIG. 1 is a block diagram showing a pedestrian detection device based on a deep learning algorithm according to the present invention. Referring to FIG. 1, the deep learning algorithm-based pedestrian detection device (100) according to the present invention may include a detection unit (110), a collection unit (120), a preprocessing unit (130), an inference unit (140), a control unit (150), and an output unit (160) on the device. The above detection unit (110) is equipped with a high-resolution camera to capture video in real time from the moment a pedestrian approaches the crosswalk until they complete crossing. The continuous detection unit (110) can acquire video information regarding a pre-set area of interest in relation to the crosswalk. In addition, the continuous detection unit (110) may include an infrared sensor, LiDAR, an ultrasonic sensor, etc. for auxiliary detection. This allows for accurate detection of pedestrians even when lighting or weather conditions are poor, thereby increasing the reliability of the data. The continuous collection unit (120) can collect image information captured by the detection unit (110). The above preprocessing unit (130) can remove noise using a digital image filter and normalize the input image through white balancing, color temperature adjustment, angle correction, distortion correction, size correction, blur correction, etc. The above inference unit (140) can receive preprocessed image information through the YOLO8m model and detect pedestrian objects in real time. The above inference unit (140) can determine whether an object existing in the image information is a person after learning in advance whether the pedestrian included in the object is a person through the YOLO8m model. The above control unit (150) can execute the overall control function of the pedestrian detection device. That is, the above control unit (150) can execute the overall control function of the pedestrian detection device using the program and data stored in the storage unit. The output unit (160) can guide the pedestrian to safely cross the road by activating a voice warning sound or illuminating an LED traffic light through the control unit (150) for the visually impaired or the elderly when a pedestrian is detected. In addition, the output unit (160) can generate a warning sound to notify drivers of vehicles passing the crosswalk of the approach of pedestrians. Although embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvem