• Korean Journal of Applied Biomechanics
• /
• v.12 no.2
• /
• pp.345-359
• /
• 2002

The purpose of the study was to provide the fundamental data to instruct athletes through the analysis athletes' movement in javelin. Three athletes in the level of national representative were participated in this study. The study analyzed kinematic variables(lead foot and releasing javelin) through 3-D analysis and obtained the following results. 1. During withdrawal, it is important to maintain of running horizontal velocity. 2. It was showed that throng average height was $84{\pm}3.3%$ and javelin adequative degree, Among the athletes, $S_2$ who had the best record was released the javelin with the fast velocity, but throw the javelin with the less releasing velocity. 3. $S_2$ released after lead foot were completely landed and therefore it is no problem in a kinematic aspect. However, $S_1$ angle was too small. it caused increase of release velocity to be prevented. 4. $S_2$ showing the best result indicated shorter in duration time. Generally, the shorter duration time in release phase showed the longer release distance. Especially $S_1$ and $S_3$ showing the worse result indicated the longer duration time in preparatory phase, causing the breakup of force. Therefore to improve the record, it should be decreased the duration time in preparatory phase. 5. Compared with $S_1$ and $S_3$, $S_2$ showing the best record indicated the higher velocity in center of mass, trunk, upper arm, lower arm and hand That is the higher velocity of upper arm at release leaded the better velocity transfer from upper arm to following lower arm and hand, these action should be considered to be helpful of better record. According to the above conclusion, when the athletic leaders cauch athletes, they should focus on maintaining knee angle, upper body and hip angle in a previous stage of release and throwing angle, throwing height, throwing velocity in a release stage.

• Korean Journal of Soil Science and Fertilizer
• /
• v.19 no.4
• /
• pp.339-344
• /
• 1986

Frost damage which can reduce yields, impair fruit quality and cause loss of trees is closely related to the occurrence of daily minimum temperature. Horizontal distribution of air temperature minima can be characterized by conditions of radiational cooling and gravitational movement of cold air, which are influenced by the regional topographic features. Observations were made on the air temperature minima over Yangpyeong area, to delineate potential effects of topography on the temperature pattern during spring season. Two routes were selected for the observation. Liquid glass minimum thermometers were installed at 17 sites through the old peach orchards which had been closed due to the frequent freeze-frost hazards during the recent years. This route was 8.5km long and the highest point was 350m above mean sea level. The other route, which was 2.5km in distance, was run with a digital resistance thermometer during the hour just before sunrise. Observations were made both on a calm-clear day (April 30, 1986) and a windy-overcast day (May 1, 1986). The temperature on April 30 was in increasing trend with elevation but this was modified at near the riverside and the downtown area. An orchard lying on a hilltop showed the temperature $1^{\circ}C$ higher than near by lowland of which elevation was about 30m lower. The minimum temperature on the overcast day was little affected by terrestrial conditions but by the atmospheric lapse condition. The peach orchards severely damaged by cold air were found in the area where the lowest minimum temperature was observed. The results may be useful for selection of the proper orchard location to be developed in an area.

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.163-177
• /
• 2019

As smartphones are getting widely used, human activity recognition (HAR) tasks for recognizing personal activities of smartphone users with multimodal data have been actively studied recently. The research area is expanding from the recognition of the simple body movement of an individual user to the recognition of low-level behavior and high-level behavior. However, HAR tasks for recognizing interaction behavior with other people, such as whether the user is accompanying or communicating with someone else, have gotten less attention so far. And previous research for recognizing interaction behavior has usually depended on audio, Bluetooth, and Wi-Fi sensors, which are vulnerable to privacy issues and require much time to collect enough data. Whereas physical sensors including accelerometer, magnetic field and gyroscope sensors are less vulnerable to privacy issues and can collect a large amount of data within a short time. In this paper, a method for detecting accompanying status based on deep learning model by only using multimodal physical sensor data, such as an accelerometer, magnetic field and gyroscope, was proposed. The accompanying status was defined as a redefinition of a part of the user interaction behavior, including whether the user is accompanying with an acquaintance at a close distance and the user is actively communicating with the acquaintance. A framework based on convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent networks for classifying accompanying and conversation was proposed. First, a data preprocessing method which consists of time synchronization of multimodal data from different physical sensors, data normalization and sequence data generation was introduced. We applied the nearest interpolation to synchronize the time of collected data from different sensors. Normalization was performed for each x, y, z axis value of the sensor data, and the sequence data was generated according to the sliding window method. Then, the sequence data became the input for CNN, where feature maps representing local dependencies of the original sequence are extracted. The CNN consisted of 3 convolutional layers and did not have a pooling layer to maintain the temporal information of the sequence data. Next, LSTM recurrent networks received the feature maps, learned long-term dependencies from them and extracted features. The LSTM recurrent networks consisted of two layers, each with 128 cells. Finally, the extracted features were used for classification by softmax classifier. The loss function of the model was cross entropy function and the weights of the model were randomly initialized on a normal distribution with an average of 0 and a standard deviation of 0.1. The model was trained using adaptive moment estimation (ADAM) optimization algorithm and the mini batch size was set to 128. We applied dropout to input values of the LSTM recurrent networks to prevent overfitting. The initial learning rate was set to 0.001, and it decreased exponentially by 0.99 at the end of each epoch training. An Android smartphone application was developed and released to collect data. We collected smartphone data for a total of 18 subjects. Using the data, the model classified accompanying and conversation by 98.74% and 98.83% accuracy each. Both the F1 score and accuracy of the model were higher than the F1 score and accuracy of the majority vote classifier, support vector machine, and deep recurrent neural network. In the future research, we will focus on more rigorous multimodal sensor data synchronization methods that minimize the time stamp differences. In addition, we will further study transfer learning method that enables transfer of trained models tailored to the training data to the evaluation data that follows a different distribution. It is expected that a model capable of exhibiting robust recognition performance against changes in data that is not considered in the model learning stage will be obtained.