• Journal of Information Processing Systems
• /
• v.19 no.6
• /
• pp.730-744
• /
• 2023

The topic of this paper is the recognition of human activities using egocentric vision, particularly captured by body-worn cameras, which could be helpful for video surveillance, automatic search and video indexing. This being the case, it could also be helpful in assistance to elderly and frail persons for revolutionizing and improving their lives. The process throws up the task of human activities recognition remaining problematic, because of the important variations, where it is realized through the use of an external device, similar to a robot, as a personal assistant. The inferred information is used both online to assist the person, and offline to support the personal assistant. With our proposed method being robust against the various factors of variability problem in action executions, the major purpose of this paper is to perform an efficient and simple recognition method from egocentric camera data only using convolutional neural network and deep learning. In terms of accuracy improvement, simulation results outperform the current state of the art by a significant margin of 61% when using egocentric camera data only, more than 44% when using egocentric camera and several stationary cameras data and more than 12% when using both inertial measurement unit (IMU) and egocentric camera data.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.17 no.6
• /
• pp.1137-1144
• /
• 2022

In this study, an artificial intelligence(AI) was developed to help with facial expression practice in order to express emotions. The developed AI used multimodal inputs consisting of sentences and facial images for deep neural networks (DNNs). The DNNs calculated similarities between the emotions predicted by the sentences and the emotions predicted by facial images. The user practiced facial expressions based on the situation given by sentences, and the AI provided the user with numerical feedback based on the similarity between the emotion predicted by sentence and the emotion predicted by facial expression. ResNet34 structure was trained on FER2013 public data to predict emotions from facial images. To predict emotions in sentences, KoBERT model was trained in transfer learning manner using the conversational speech dataset for emotion classification opened to the public by AIHub. The DNN that predicts emotions from the facial images demonstrated 65% accuracy, which is comparable to human emotional classification ability. The DNN that predicts emotions from the sentences achieved 90% accuracy. The performance of the developed AI was evaluated through experiments with changing facial expressions in which an ordinary person was participated.

• Journal of Institute of Control, Robotics and Systems
• /
• v.21 no.6
• /
• pp.503-509
• /
• 2015

In this paper, we designed and tested an ankle joint mechanism for a gait rehabilitation robot. Gait rehabilitation programs are designed to improve the natural leg motion of patients who have lost their walking capabilities by accident or disease. Strengthening the muscles of the lower-limbs and stimulation of the nervous system corresponding to walking helps patients to walk again using gait assistive devices. It is an obvious requirement that the rehabilitation system's motion should be similar to and as natural as the normal gait. However, the system being used for gait rehabilitation does not pay much attention to ankle joints, which play an important role in correct walking as the motion of the ankle should reflect the movement of the center of gravity (COG) of the body. Consequently, we have designed an ankle mechanism that ensures the safety of the patient as well as efficient gait training. Also, even patients with low leg muscle strength are able to operate the ankle joint due to the direct-drive mechanism without a reducer. This safety feature prevents any possible adverse load on the human ankle. The additional degree of freedom for the roll motion achieves a gait pattern which is similar to the normal gait and with a greater degree of comfort.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37 no.8A
• /
• pp.639-647
• /
• 2012

Vision-based on-road vehicle detection is one of the key techniques in automotive driver assistance systems. However, due to the huge within-class variability in vehicle appearance and environmental changes, it remains a challenging task to develop an accurate and reliable detection system. In general, a vehicle detection system consists of two steps. The candidate locations of vehicles are found in the Hypothesis Generation (HG) step, and the detected locations in the HG step are verified in the Hypothesis Verification (HV) step. Since the final decision is made in the HV step, the HV step is crucial for accurate detection. In this paper, we propose using a reduced multivariate polynomial pattern classifier (RM) for the HV step. Our experimental results show that the RM classifier outperforms the well-known Support Vector Machine (SVM) classifier, particularly in terms of the fast decision speed, which is suitable for real-time implementation.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.18 no.6
• /
• pp.243-250
• /
• 2018

This paper designs and implements a smart care system for the senior citizen who lives alone. Recently, as the level of living has increased due to the rapid improvement of medicine, living standard and environment, the proportion of the elderly population is increasing. In addition, the proportion of the elderly living alone, which is increasing with the aging society, suggests that the provision of services such as the elder care system and emergency notification is becoming an important issue. However, since the existing emergency notification technology analyzes fixed CCTV images, it is difficult to monitor in the blind spot of CCTV and to move to a place where the camera is not installed. There is a problem that it can not be performed. Therefore, in this paper, we design and develop a smart care system that utilizes robot and object tracking technology that can move in real time to overcome these shortcomings. This enables real-time monitoring regardless of the location, and prompts for assistance in case of an emergency, so that it can provide convenience to cares and assistants.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.7
• /
• pp.849-856
• /
• 2020

Recently, as the number of high-height vehicles such as double-decker buses has increased, collision accidents have occurred in bridges and tunnels due to the deviation from the designated routes and driver's carelessness. In the case of the existing front collision warning system, it is limited to vehicles and pedestrians, so it is difficult to use it as a pass height warning system for the high height vehicles. In this paper, we propose a system that generates a warning by determining the correlation and time series characteristics of data for each segment using multiple lidar sensors and then determining the possibility of collision in the upper part of the vehicle. Also, the proposed system confirmed the proper operation through a real-time driving test and a system performance evaluation by the Korea Automobile Testing & Research Institute.

• Journal of Broadcast Engineering
• /
• v.28 no.2
• /
• pp.178-184
• /
• 2023

Finding texts in general scene images and recognizing their contents is a very important task that can be used as a basis for robot vision, visual assistance, and so on. However, for the low-resolution text images, the degradations, such as noise or blur included in text images, are more noticeable, which leads to severe performance degradation of text recognition accuracy. In this paper, we propose a new Korean text image super-resolution based on a Transformer-based model, which generally shows higher performance than convolutional neural networks. In the experiments, we show that text recognition accuracy for Korean text images can be improved when our proposed text image super-resolution method is used. We also propose a new Korean text image dataset for training our model, which contains massive HR-LR Korean text image pairs.

• Journal of Venture Innovation
• /
• v.5 no.4
• /
• pp.179-189
• /
• 2022

In the face of revolutionary changes in industry, the relationship between labor and income needs to be reconceptualized in the period of social revolution. The absolute decrease in labor due to the absence of labor is caused by automation, smartization, AI, robot labor, etc., which we must accept whether we want to or not. However, while gross social product and capital of the state or society increase, individual income is likely to decrease. During this transformation period, the state or politics must prepare for the problems caused by the decline in individual income. Until now, there have been various levels of discussion on social welfare or social security from the perspective of welfare or assistance. Attempts or studies at the experimental level have been conducted at the level of many countries or local governments and have found positive and negative effects. There is no basic income system that is widely implemented at the national level, and various discussions are taking place from a future-oriented perspective. Therefore, I propose to look at it from a new perspective based on the perspective so far. We explored that it is part of a positive approach to examine the importance and necessity of basic income in terms of working hours, quality of labor, income, quality of life, value of spare time, and work-life balance. The goal is to actively accept the absolute lack of working hours, replacement of mechanical labor, and polarization due to changes in the industry paradigm, and to look at the problems that come from a positive perspective. If we are going to accept it anyway, we should not look at these issues as short-sighted, but prepare them preemptively and establish a primitive plan from a long-term and overall perspective. Smartphones have changed the world over the past decade and have been lost, but wouldn't there be a lot of new discoveries? Shouldn't we think of it as a great opportunity to improve the quality of life through technological changes?