• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.150-153
• /
• 2022

This paper reports a technique that automatically extracts object shapes through Dense-Net, and subsequently, detects the objects using Mask R-CNN in a manufacturing site, in which workers and objects are mixed. It is based on the customized factory dataset by targeting workers, machines, tools, control boxes, and products as the objects. Mask R-CNN supports multi-object recognition as a well-known object recognition method, while Dense-Net effectively extracts a feature from multiple and overlapping objects. After immediate implementation using the two technologies, the object is naturally extracted from a still image of the manufacturing site to describe image. Afterwards, the result is planned to be used to detect workers' abnormal behavior by adding a label on the objects.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.213-215
• /
• 2022

Recently, deep learning technology that processes images has been widely used in fire detection, autonomous driving, and defective product detection. In particular, in order to determine whether a product is contaminated or not, it can be identified through the contaminants passed from the existing sensor data, but technologies for recognizing cracks in products or contaminants themselves as images are being actively studied in various fields. In this paper, a system for classifying uncontaminated normal cups and contaminated cups through images was designed and implemented. The image was analyzed using an open image and a photographed image, and the image was analyzed by extracting the upper part of the cup image using Google Objectron for 3D object recognition. Through this study, it is thought that it will be used in various ways for research that can extract the contamination level of products required in the hygiene field based on images.

• Smart Media Journal
• /
• v.13 no.5
• /
• pp.19-25
• /
• 2024

The biggest type of behavior that prevents people with developmental disabilities from entering society is aggressive behavior. Aggressive behavior can pose a threat not only to the personal safety of the person with a developmental disability, but also to the physical safety of others. In this study, we propose a wearable system using a low-power processor. The proposed system uses an IMU (Inertial Measurement Unit) to analyze user behavior, and when attack behavior is not detected for a certain period of time through an LED array attached to the developed system, an interesting LED is displayed. By expressing patterns, we provide behavioral intervention through compensation to people with developmental disabilities. In order to implement a system that must be worn for a long time in a power-limited environment, we present a method to optimize performance and energy consumption across all stages, from data preprocessing to AI model application.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.48 no.2
• /
• pp.96-105
• /
• 2011

Ubiquitos sensor network(USN) is a technology which is widely used in our life. This paper introduces an example of design and implementation for a system which is based on the USN technology and can provide an efficient management tool for a space that should be precisely controlled for a certain range of uniformity in temperature and humidity. This introduced system builds a wireless sensor network using a number of sensor modules that are equipped with temperature and humidity sensors, and collects temperature and humidity information in real-time while simultaneously providing a method for monitoring the status of temperature and humidity by the graphical user interface. Also, the system will give a warning signal if the monitored values are differ from the pre-specified values of temperature and humidity for each sensor module more than a certain amount of tolerance. This temperature and humidity logging and monitoring system can perform better management for the space easily and efficiently by automating the existing manual method for data collection and management. Furthermore, using the stored data, it can make possible to perform post-analysis on the problems caused by temperature and humidity and to obtain information for environmental enhancement for the space.

• Journal of rehabilitation welfare engineering & assistive technology
• /
• v.11 no.2
• /
• pp.115-124
• /
• 2017

This study was to develop the balance insole system for detecting and improving the muscle imbalance of left and right side in lower limbs. We were to verify the validation of balance insole system by analyzing the strategy of muscular activities and foot pressure according to sound feedback. We developed the balance insole based FSR sensor modules for estimating the muscle imbalance using detecting foot pressure. The insole system was FPCB have 8-spot FSR sensor with sensitivity range of 64-level. The participants were twenty peoples who have muscle strength differences in left and right legs over 20%. We measured the muscular activity and foot pressure of left and right side of lower limbs in various gait environment for verifying the improvement effect of muscle imbalance according to sound feedback. They performed gait in slope at 0, 5, 10, 15% and velocity at 3, 4, 5km/h. The result showed that the level of muscle imbalance reduced within 30% for sound feedback of balance insole system contrast to high level of muscle imbalance at 169.9~246.8% during normal gait for increasing slope and velocity. This study found the validation of balance insole system with sound feedback stimulus. Also, we thought that it is necessary to research on the sensitivity of foot area, detection of muscle imbalance and processing algorithm of correction threshold spot.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.38C no.7
• /
• pp.630-642
• /
• 2013

To transmit vital signal stream of mobile patients remotely, it requires mobility of patient and watcher, sensing function of patient's abnormal symptom and self-organizing service binding of related computing resources. In the existing relative researches, the vital signal stream is transmitted as a centralized approach which exposure the single point of failure itself and incur data traffic to central server although it is localized service. Self-organizing middleware platform based on heterogenous overlay network is a middleware platform which can transmit real-time data from sensor device(including vital signal measure devices) to Smartphone, TV, PC and external system through overlay network applied self-organizing mechanism. It can transmit and save vital signal stream from sensor device autonomically without arbitration of management server and several receiving devices can simultaneously receive and display through interaction of nodes in real-time.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.11
• /
• pp.205-210
• /
• 2014

This is concerned with the technology to display the design on Porcelain and adjust malfunction for moving effect and light intensity by curator. More precisely, the technology makes it possible that the porcelain is connected to Light module which is the device for controlling light emission and rotating rolling plate, etc that are connected to LED light module, optical fiber and controller that is for scenario from the given storytelling. In addition, with a WiFi portable device (Smart-phone, other mobile device). equipped with a scenario programs, information for operation, failure and malfunction can be obtained and analyzed in real-time, and menu color and alarm is alerted when the displaying design is in abnormal status, which makes the early reactions to the status. Furthermore, the collected data can be sent through WiFi network to the device and PC managed by the curator specialized in managing the design on the Porcelain, thus the technology could help the curator who have less knowledge about moving pattern on the Porcelain. There is always a possibility of malfunction due to various condition that are caused by wring-harness when modules are wired-connected. In this research, in order to overcome this problem, we propose a system configuration that can do monitoring and diagnosis with a device for collecting data from LED control module, Light emission sensor and a personal WiFi device. Also, we performed connection between optical Fiber and LED and interlock for the system defined by the definition for information and storytelling scenario.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.10
• /
• pp.27-35
• /
• 2023

In this paper, we propose a process of increasing productivity by applying a deep learning-based defect detection and classification system to the prepreg fiber manufacturing process, which is in high demand in the field of producing composite materials. In order to apply it to toe prepreg manufacturing equipment that requires a solution due to the occurrence of a large amount of defects in various conditions, the optimal environment was first established by selecting cameras and lights necessary for defect detection and classification model production. In addition, data necessary for the production of multiple classification models were collected and labeled according to normal and defective conditions. The multi-classification model is made based on CNN and applies pre-learning models such as VGGNet, MobileNet, ResNet, etc. to compare performance and identify improvement directions with accuracy and loss graphs. Data augmentation and dropout techniques were applied to identify and improve overfitting problems as major problems. In order to evaluate the performance of the model, a performance evaluation was conducted using the confusion matrix as a performance indicator, and the performance of more than 99% was confirmed. In addition, it checks the classification results for images acquired in real time by applying them to the actual process to check whether the discrimination values are accurately derived.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.33 no.6
• /
• pp.1099-1110
• /
• 2023

The recognition system of autonomous driving and robot navigation performs vision work such as object recognition, tracking, and lane detection after multi-sensor fusion to improve performance. Currently, research on a deep learning model based on the fusion of a camera and a lidar sensor is being actively conducted. However, deep learning models are vulnerable to adversarial attacks through modulation of input data. Attacks on the existing multi-sensor-based autonomous driving recognition system are focused on inducing obstacle detection by lowering the confidence score of the object recognition model.However, there is a limitation that an attack is possible only in the target model. In the case of attacks on the sensor fusion stage, errors in vision work after fusion can be cascaded, and this risk needs to be considered. In addition, an attack on LIDAR's point cloud data, which is difficult to judge visually, makes it difficult to determine whether it is an attack. In this study, image scaling-based camera-lidar We propose an attack method that reduces the accuracy of LCCNet, a fusion model (camera-LiDAR calibration model). The proposed method is to perform a scaling attack on the point of the input lidar. As a result of conducting an attack performance experiment by size with a scaling algorithm, an average of more than 77% of fusion errors were caused.

• Korean Journal of Agricultural and Forest Meteorology
• /
• v.25 no.3
• /
• pp.182-196
• /
• 2023

Accurate and timely estimation of crop yields is crucial for various purposes, including global food security planning and agricultural policy development. Remote sensing techniques, particularly using vegetation indices (VIs), have show n promise in monitoring and predicting crop conditions. However, traditional VIs such as the normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) have limitations in capturing rapid changes in vegetation photosynthesis and may not accurately represent crop productivity. An alternative vegetation index, the near-infrared reflectance of vegetation (NIRv), has been proposed as a better predictor of crop yield due to its strong correlation with gross primary productivity (GPP) and its ability to untangle confounding effects in canopies. In this study, we investigated the potential of NIRv in estimating crop yield, specifically for corn and soybean crops in major crop-producing regions in 14 states of the United States. Our results demonstrated a significant correlation between the peak value of NIRv and crop yield/area for both corn and soybean. The correlation w as slightly stronger for soybean than for corn. Moreover, most of the target states exhibited a notable relationship between NIRv peak and yield, with consistent slopes across different states. Furthermore, we observed a distinct pattern in the yearly data, where most values were closely clustered together. However, the year 2012 stood out as an outlier in several states, suggesting unique crop conditions during that period. Based on the established relationships between NIRv peak and yield, we predicted crop yield data for 2022 and evaluated the accuracy of the predictions using the Root Mean Square Percentage Error (RMSPE). Our findings indicate the potential of NIRv peak in estimating crop yield at the county level, with varying accuracy across different counties.