Rudy Hartanto

Real time hand gesture movements tracking and recognizing system

Human computer interaction has a long history to become more intuitive. For human being, gesture of different kind is one of the most intuitive and common communication. However, vision-based hand gesture recognition is a challenging problem, which is involved complex computation, due to high degree of freedom in human hand. In this paper, we use hand gesture captured by web-cam instead of mice, for natural and intuitive human-computer interaction. Skin detections method is used to create a segmented hand image and to differentiate with the background. A contours and convex hull algorithm is used to recognize hand area as well as the number of fingertip of hand gesture image to be mapped with button. Moreover, for detection of hand gesture motion, we use Lucas-Kanade pyramidal algorithm. The result shows that this system can operate well so we can interact with computer using our hand gesture instead using a mouse.

Preliminary design of static indonesian sign language recognition system

Human computer interaction has a long history to become more intuitive. For human being, especially for the deaf, gesture of different kind is one of the most intuitive and common communication. In this paper we focus on creating a system to identified and translate hand gesture pose to Indonesian alphabets. Skin detections method is used to create a segmented hand image and to differentiate with the background. A contours is used to localize hand area. SIFT algorithm in advanced, were used to recognize the signed gesture. The result shows that this system can operate well in translated hand gesture image of sign into Indonesian alphabets.

Real time static hand gesture recognition system prototype for Indonesian sign language

Sign language uses gestures instead of speech sound to communicate. However, it is rare that the normal people try to learn the sign language for interacting with deaf people. Therefore, the need for a translation from sign language to written or oral language becomes important. In this paper, we propose a prototype system that can recognize the hand gesture sign language in real time. We use HSV (Hue Saturation Value) color space combined with skin detection to remove the complex background and create segmented images. Then a contour detection is applied to localize and save hand area. Further, we use SURF algorithm to detect and extract key point features and recognize each hand gesture sign alphabet by comparing with these user image database. Based on the experiments, the system is capable to recognize hand gesture sign and translate to Alphabets, with recognize rate 63 % in average.

Android based real-time static Indonesian sign language recognition system prototype

Sign language uses gestures instead of speech sounds to communicate. But in general, normal people rarely trying to learn sign language to interact with the deaf community. Recently, there are many sign language recognition system that had been developed. But most of them were implemented using desktop and laptop computer, which is impractical due to its weight and size. This paper presents a prototype of real time static Indonesian sign language recognition using the Android Smart Phone, so it can be used anywhere and anytime. YCrCb color space combined with skin color detection is used to remove the background image and form a segmented image. Detection contour with convex hull algorithms are used to localize and save an area of the hand. Convexity defect algorithms then are used to extract the hand gestures features using a radiant line from the center of the palm to the fingertips. The classification of hand gesture that performs sign alphabets is accomplished using back propagation neural network algorithm in order to determine a suitable alphabet. The performance test of the system is done by recognizing some variation hand gesture poses for Indonesian sign language alphabet. The results show that the system can detect the position of the users hand. Furthermore, the system can recognize the alphabet sign from user hand gesture input, reaching 91.66% success rate in testing using Android devices in real time.

Implementation of Augmented Reality Technology in Sangiran Museum with Vuforia

Archaeological object is an evidence of life on ancient relics which has a lifespan of millions years ago. The discovery of this ancient object by the Museum Sangiran then is preserved and protected from potential damage. This research will develop Augmented Reality application for the museum that display a virtual information from ancient object on display. The content includes information as text, audio, and animation of 3D model as a representation of the ancient object. This study emphasizes the 3D Markerless recognition process by using Vuforia Augmented Reality (AR) system so that visitor can access the exhibition objects through different viewpoints. Based on the test result, by registering image target with 25o angle interval, 3D markerless keypoint feature can be detected with different viewpoint. The device must meet minimal specifications of Dual Core 1.2 GHz processor, GPU Power VR SG5X, 8 MP auto focus camera and 1 GB of memory to run the application. The average success of the AR application detects object in museum exhibition to 3D Markerless with a single view by 40%, Markerless multiview by 86% (for angle 0° - 180°) and 100% (for angle 0° - 360°). Application detection distance is between 23 cm and up to 540 cm with the response time to detect 3D Markerless has 12 seconds in average.

A comparison of different part-of-speech tagging technique for text in Bahasa Indonesia

Part of speech tagging has some different methods or techniques to the problem in assigning each word of a text with a part-of-speech tag. In this paper, we conducted some part-of-speech tagging techniques for Bahasa Indonesia experiments using statistical approach (Unigram, Hidden Markov Models) and Brills tagger. In this study, we used Supervised POS Tagging approach requiring a large number of annotated training corpuses to tag properly. We used some resource annotation corpus of Bahasa. Those corpuses were implemented with POS Tagging techniques. We subsequently compared and analyzed the results. We also compared the accuracy and highlighted some advantages and disadvantages for every technique we used. Unigram showed a higher accuracy compared to HMM and Brill tagger with 88,37% on a tagged corpus.

High-resolution automated Fugl-Meyer Assessment using sensor data and regression model

Motor function assessment is a critical component in post-stroke rehabilitation program. Fugl-Meyer Assessment (FMA) is regarded as the most comprehensive tool to describe post-stroke patients motor function and is widely utilized. However, the FMA scoring system classifies patients motor function within only three levels (0, 1, or 2) for each assessment item. Hence, its difficult to observe minor improvement in patients motor function. To address this issue, a high-resolution scoring system was proposed in this research. Six regression models were built for six upper-extremity FMA items. These models were trained using data gathered from Microsoft Kinect sensor and glove sensor. The trained models could predict FMA scores with the resolution of 14 fractional digits. The representativeness of the predicted scores was evaluated by calculating their Pearsons correlation with the actual kinematic variables. All predicted scores represent patients motor function better than standard FMA score. In addition, the highest correlations mean of 0.58 was found on shoulder elevation item which utilized Neural Network Regression algorithm. This algorithm also outperformed the other in most assessment items. The ability to observe the changes in patients motor function in detail helps therapists providing more responsive treatment and could likely increase the patients adherence toward the given treatment.

Indonesian Automatic Speech Recognition system using CMUSphinx toolkit and limited dataset

Building Automatic Speech Recognition (ASR) needs acoustic model, language model and dictionary for intended language, which is also applied for Indonesian ASR. In this paper, Indonesian ASR was built using CMUSphinx toolkit (a Hidden Markov Model based ASR tool) with limited dataset. We use digit corpus and own made language model to trained with the limited dataset. We also investigated the implementation of trained acoustic model by examine it in different SNR condition to several people. The best achievement of word error accuracy of the acoustic model is 86% on average. By examine it in different SNR condition, we got maximum accuracy of 80% on 27.764 dB environment.

Development and evaluation of a low cost music based exergame using microsoft kinect

Non-communicable Diseases (NCD) has become a worldwide problem. In 2012, 38 million or approximately 68% of deaths in the world are caused by NCD. The impacts caused by NCD are growing year after year. Physical inactivity is addressed as one of the main factors of NCD. Numerous efforts have been made to suppress its prevalence. Providing active game (exergame) is considered to be one of the most effective solutions. Even though there are some exergames that have been developed, there are still rooms for improvement. Thus, a new exergame called Beat-Beat Fitness (BBF) will be developed in this research. BBF contains several innovative features that likely enhance the psychophysiological impact to the players. The psychophysiological effect is measured using a questionnaire to observe the improvement created by BBF compared to several existing exergame. As the result, BBF has higher level of engagement than Kinect Dance Central. It also has higher usability level than Kinect Just Dance and Kinect Dance Central. Additional features proposed in this study are proven in increasing the psychophysiological effect of an exergame. The results of this study can be used as a reference for the development of incoming exergames.

EFEK MODERASI KESUKARELAAN TERHADAP PEMBELAJARAN ONLINE PADA UNIVERSITAS SAHID SURAKARTA

E-learning is an IT tool that facilitates learning process. In order to optimize its function, e-learning need to be evaluated. There are many indicators on evaluating e-learning such as voluntariness of use. This study aims to evaluate factors that impact the usage of an e-learning system by considering voluntariness of use as moderating factor on e-learning usage. This study uses the UTAUT model that consists of four independent variables (performance expectancy, effort expectancy, social influence, and facilitating condition), two dependent variables (behavioral intention, use behavior), and one moderating variable (voluntariness). Voluntariness is hypothesized to affect the interaction of social influence and behavioral intention. This study uses Structural Equation Modeling (SEM) to validate the concepts and theories about e-learning usage’s factors at Universitas Sahid Surakarta and Moderated Structural Equation Modeling (MSEM) to validate the moderation effect of voluntariness to the interaction of social influence and behavioral intention. From SEM analysis, we can conclude that performance expectancy and effort expectancy positively influence the behavioral intention of e-learning. Meanwhile, from MSEM analysis we can conclude that voluntariness can’t moderate the usage of e-learning. Keywords: e-learning; SEM; UTAUT; voluntariness.