|
A Comparative Study of Foreground Detection using Gaussian Mixture Models- Novice to Novel |
|
Ajmal Shahbaz, Kang-Hyun Jo(University of Ulsan, Korea), Laksono Kurnianggoro(Intelligent Systems Lab., Korea) |
| Foreground detection is the classical computer vision task of segmenting out motion information from a
particular scene. Foreground detection using Gaussian Mixture Models (GMM) is the famous choice. Since first time
proposed, many researchers tried to improve GMM. This paper focuses on the comparative evaluation of three most
famous improvements in the algorithm. The improved methods are compared both qualitatively and quantitatively using
standard datasets available online. |
|
|
PID Compensating Model for Design of Ship’s Autopilot |
|
Youngkuk Kwon(Busan National Univ., Korea), Seon-Ho Park, Jang Myung Lee(Pusan National University, Korea) |
| This paper proposes an autopilot system using a PID compensating model to satisfy performance required for
the automatic navigation of ships under various marine circumstances. |
|
|
Performance Evaluation of Tactile Sensor-based Grip Control using 3D-printed Flexible Tactile Sensor |
|
Ju Kyoung Lee, Hyeong-Jun Kim, Suk Lee(Pusan National University, Korea), Kyung Chang Lee(Pukyong National University, Korea) |
| 3D-printed flexible tactile-sensor-based robotic gripper system for object grasping and experimentally verified the system performance. These flexible tactile sensors are based on pressure-sensing materials that allow pressure to be measured according to resistance change that in turn results from changes in material size because of compressive force. The sensing material consists of a mixture of multiwalled carbon nanotubes (MWCNTs) and TangoPlus, which gives it flexibility and elasticity. The tactile sensors used in this study were designed in the form of array structures composed of many li |
|
|
Region Segmentation and Classification Using Convolutional Neural Network |
|
Jang Sik Park, Hyun Gon Kang, Jong Kwan Song, Byung Woo Yoon(Kyungsung University, Korea) |
| Recent interest in the technology for recognizing the current situation, such as loitering, intrusion, abandonment of undetermined pedestrians on the CCTV images is increasing. In this paper, in order to apply the situation recognition system, authors developed a system to detect the geographic features such as roads, fences that exist in the CCTV image. When detecting the region of interest in the image, it can be easily and accurately determined. Authors used the convolutional neural network (CNN) for detecting the region. CNN, which consists of a number of hidden layers that can fully take |
|
|
A Video Based People Counting Using Detection and Tracking Algorithm |
|
Jang Sik Park, Baris H Baydargil(Kyungsung University, Korea) |
| Nowadays, people counting have a wide area of applications. The process is comprised of two stages; first, the human detection and the second, human tracking. This paper’s main objective is to combine human detection with AdaBoost using Haar-like features for a full body detection, and LBP (Local Binary Patterns) for head detection, and afterwards the target is then tracked with PDAF (Probabilistic Data Association Filter). And at last, using the virtual (visible or invisible) line on the screen, targets that cross the line are counted. The experimental system is implemented and the results ar |
|
|
Video Based Gender Recognition Using Convolutional Neural Network In Night Time |
|
Jang Sik Park, Omer Faruk Ince(Kyungsung University, Turkey) |
| Several approaches have been used for gender recognition cases. Convolutional Neural Network(CNN) is becoming the most popular approach to classify for many problems. This paper proposes night time gender recognition using CNN. Adaboost algorithm is used for people’s full body detection, local binary pattern(LBP) is performed to detect head of person, and probabilistic data association filter(PDAF) is used for pedestrian tracking. In the end, to make more accurate classification with less computation, CNN algorithm is used to distinguish detected person according to its gender. As a result of |
|