Garuda - Garba Rujukan Digital

Penerapan Desain User Experience Pada Aplikasi Penghitungan Matematika Bagi Anak Penyandang Tunagrahita di Quali International Surabaya Yohanes Hendra Wijaya; Liliana Liliana; Lily Eka Sari
Jurnal Infra Vol 9, No 2 (2021)
Publisher : Jurnal Infra

Show Abstract | Download Original | Original Source | Check in Google Scholar

Intellectual disability is a condition that happens to a child under 18 years old. This condition can lead to cognitive problems and below average intelligence. Because of this condition, those children need special care, especially for learning. One of the important lessons to learn is mathematics, because they need it to live in society. This research applies a design that focused on children with intellectual disability in an android application to learn mathematics. The design uses a simple, attractive, and interactive display so they don’t get confused or get bored easily. The materials used are adjusted to the competency standards for those children to suit their abilities. The application is designed to be an alternative teaching aid in mathematics. The results of this study are that children with intellectual disability can use 70.5% of the features of this application which are adjusted to their abilities. The application succeeded in helping children to improve their skills. According to the teachers at Quali International Surabaya, the material used in this application is in accordance with the needs of the children, and is delivered quite easily and clearly. Application is also suitable to be used as a teaching aid for mathematics lessons at school.

Meningkatkan Variasi Tindakan Non-Playable Character Pada Game Survival Menggunakan Metode Markov Hendra Winata; Liliana Liliana; Hans Juwiantho
Jurnal Infra Vol 9, No 2 (2021)
Publisher : Jurnal Infra

Show Abstract | Download Original | Original Source | Check in Google Scholar

Digital games or often called Video games are common today. The development of game variants makes games never stop improving, especially in the Artificial Intelligence section. Each game has its own artificial intelligence so that many variations are generated and make a game unique. This research tries to make a variation of the actions taken by NPCs against players. In an effort to make these variations, the Markov Chain method is used to help state selection. Markov Chain method is combined with Finite-State Machine for NPC state selection. Based on the results of testing and questionnaires, 80.4% strongly agree and 19.6% agree that the resulting NPC has a large variety of actions. The results of the questionnaire also found that 69.6% were very unrealistic and 30.4% said that NPCs were unrealistic or did not imitate human behavior.

Sistem Penunjang Belanja Pedagang Keliling Di Lokasi Sekitar Menggunakan Haversine Berbasis Android Richard Gozali; Liliana Liliana; Yulia Yulia
Jurnal Infra Vol 9, No 2 (2021)
Publisher : Jurnal Infra

Show Abstract | Download Original | Original Source | Check in Google Scholar

A traveling merchant is an effort made by a person to sellgoods/services such as vegetables, fruit, and others by travelingto places with certain routes. During this pandemic, some peoplebuy products online because people are afraid to go to placeswith large crowds. This surge in demand has made many peoplelook for sellers of vegetables and fruits. With this research, it canhelp customers to determine the location of the existence oftraveling merchants in the vicinity.By using the haversine formula, you can find out the approximatedistance between the customer and the traveling merchant.During the pandemic and the habit of staying at home, manypeople prefer to cook for themselves compared to buying. It'shard to decide which dish to cook. With the use of beautifulsoup,you can take the recipe menu of the food you want to make.From the results of the survey given to customers and travelingmerchants, it has been shown that the application is made easy touse and the survey results provide a satisfactory value. Based ontesting the distance between haversine and google maps. Theresulting haversine formula is quite accurate when compared togoogle maps. The results of the comparison of the averagedistance produced are 0.315 Km.

Finite State Machine untuk Menghasilkan Kalimat Sebagai Alat Bantu Komunikasi Bagi Anak Penyandang Autisme Steven Hans Gunadi Chua; Liliana Liliana; Lily Eka Sari
Jurnal Infra Vol 9, No 2 (2021)
Publisher : Jurnal Infra

Show Abstract | Download Original | Original Source | Check in Google Scholar

One of the obstacles for children with autism is communication. Children with autism, who are nonverbal, have difficulty interacting with other people or just expressing their needs. When the child's wishes are not fulfilled, the child with autism will tend to throw a tantrum, which may be harmful because it can hurt the child. Currently, there is intervention by using the PECS method that can help children with autism learn to communicate by using picture cards. However, the drawback of this PECS method is the large number of picture cards that must be provided for children with autism.In this study, an Android application was created as a communication tools to children with autism learn to communicate. The application follows the stages of using PECS and also utilizes the finite state machine to generate sentences.Based on the results of this study, the application was able to help children with autism, who are nonverbal to communicate, although some children need more time to understand how to use the application.

Penerapan Metode Goal Oriented Action Planning untuk Agent AI pada Turn Based Tactics Video Game Ryan Chandra Kusuma; Liliana Liliana; Hans Juwiantho
Jurnal Infra Vol 9, No 2 (2021)
Publisher : Jurnal Infra

Show Abstract | Download Original | Original Source | Check in Google Scholar

In turn-based tactics game, difficulties often placed on resources owned by enemies. Players have to do repetitive action to counterbalance enemy’s resources. To make players spent more time on strategies rather than counterbalance enemy’s resources, goal-oriented action planning will be implemented for AI. It’s expected that AI GOAP even without extra resources can replace AI FSM with extra resources.Goal-Oriented Action Planning (GOAP) is a decision-making method that capable of making a character not only do what it will do, but also determine how to do it. A* is a method that looks for a path by exploring the minimum number of nodes with minimum cost solution. This research combines GOAP and A* search. GOAP in this research has several variations of actions based on health points. Result of the research shows that AI GOAP without extra resources has 33.33% winrate against AI FSM with extra resources, and 86.66% against AI FSM with extra resources but reduced power unit. The results of respondents from various players with different experiences show that the difficulty of AI FSM with extra resources is higher, the level of player satisfaction and AI’s realistic level is higher when fought against AI GOAP without extra resources.

Voice Alert Sebagai Alat Bantu Penglihatan di Lingkungan Rumah dan Jalanan Secara Umum Berbasis Android Kevin Christian Salim; Liliana Liliana; Rolly Intan
Jurnal Infra Vol 9, No 1 (2021)
Publisher : Jurnal Infra

Show Abstract | Download Original | Original Source | Check in Google Scholar

According to research conducted by the Governors Highway Safety Assocation in 2017 showed that there are about 6 thousand pedestrians killed in America due to the habit of using smartphones while walking. Another study from Jeff Ronsen showed that the brain is overloaded and cannot function properly when performing these two activities at the same time. Walking while using a smartphone makes the concentration on the atmosphere of the road split and makes pedestrians not focus on the road but rather on the smartphone. Such behavior results in an increased risk of pedestrian accidents. One solution to prevent accidents above is to use the smartphone camera to take pictures in front of the user.Smartphone cameras can be used to retrieve input data in the form of images in real time which is then carried out the process of object detection and issue alerts in the form of sounds that mention the name of the detected object to smartphone users. Detected objects are objects that are generally located in home and street environments such as humans, cars, bicycles, motorcycles, and stop signs. Object detection using SSD MobileNet applied transfer learning that is further trained by using google open image dataset v6 dataset. The result of transfer learning is weight used to detect objects from android camera input.The test results showed that SSD_MobileNet_V2 with a learning rate of 0.01 and steps 10,000 has the best mAP value with 80% in detecting objects. The SSD_MobileNet_V2 can detect objects with an inference time speed of 80ms – 110ms in real time in a standby device, and voice alerts by instantly issuing alerts when an object is detected.

Deteksi Aktivitas Manusia Berdasarkan Data Skeleton dengan Menggunakan Modifikasi VGG16 Daniel Subroto; Liliana Liliana
Jurnal Infra Vol 9, No 1 (2021)
Publisher : Jurnal Infra

Show Abstract | Download Original | Original Source | Check in Google Scholar

In general, detection of human activity is carried out to detect activity in daily life. With further development, the detection of human activity is utilized to detect suspected activity (not routine) as an early warning application. Detection of human activity in early warning applications will then be implemented in other systems. However, there are several problems in detecting human activity, among others, the presence of variations in performing an activity, the movement of transitions between activities, and the similarity of movements in different activities. Detection of human activities with existing variations can be done if utilizing a deep learning approach to conduct the training process. The deep learning method used is VGG16. VGG16 will receive input in the form of skeleton data images. The skeleton data used is obtained from the NTU RGB+D dataset. Skeleton data will be represented as 2D images by going through a process of covering the crop, converted into grayscale, resizing, and connecting 10 images into 10 channels for each sequence of activity sequences. To detect human activity is applied transfer learning on VGG16 that is changing the fully connected layer. VGG16 modification test results with skeleton data representation resulting in the highest accuracy rate of 54.59%. This level of accuracy is obtained from model testing using the same dataset as the training dataset. The VGG16 modification is still the best model based on testing with other Convolutional Neural Network models. Modification of VGG16 can classify indoor activities.

Dynamics Difficulty Adjusment Metode Evolutionary MCTS with Flexible Search Horizon pada Multi-Action Adversarial Games untuk Penyesuaian Tingkat Permainan Andhika Evantia Irawan; Liliana Liliana; Hans Juwiantho
Jurnal Infra Vol 9, No 1 (2021)
Publisher : Jurnal Infra

Show Abstract | Download Original | Original Source | Check in Google Scholar

Dynamic Difficulty Adjustment (DDA) is a method that modifies AI behavior to suit the player's abilities. So far, research on DDA in Monte Carlo Tree Search has been able to provide an appropriate level of challenge. However, the advantages of MCTS in finding solutions to long-term strategies have not been maximally implemented because so far it is only used in 2D real-time fighting games, which are short-term strategy game.This study combines DDA with evolutionary monte carlo tree search with flexible horizon (FH-EMCTS). FH-EMCTS is combination of vanilla MCTS with an Evolutionary algorithm. This method increases the length of the search space to certain extent. Giving DDA to FH-EMCTS is done by changing the way of selecting actions and assessing each node.The result of this research is that AI agents that use FH-EMCTS with DDA can be implemented into multi-action adversarial game and can provide balanced level of difficulty to other AI agents and humans. Based on the results of survey of AI agents against humans, it shows that the most fun and realistic AI agents are not the AI agents who have the best ability of winning percentage but AI agents who have win rate of around 50%.

Penerapan Convolutional Neural Network Untuk Klasifikasi Kanker Kulit Melanoma Pada Dataset Gambar Kulit Michael Kurniawan Soegeng; Liliana Liliana; Agustinus Noertjahyana
Jurnal Infra Vol 9, No 1 (2021)
Publisher : Jurnal Infra

Show Abstract | Download Original | Original Source | Check in Google Scholar

Melanoma skin cancer is one of the most dangerous skin cancers where the ferocity and speed of metastasis has caused a high mortality rate among afflicted when the cancer is not treated. Early detection of the cancer and prevention by removing the affected skin have been shown to decrease the mortality rate on afflicted patient. Thus, development of a method to help automatically diagnose the cancer and classify between cancer and normal mole or birthmark is needed. Previous methods still show limitations in classifying melanoma skin cancer. This study proposes a classification system using convolutional neural network trained on the original ISIC 2020 dataset and hair removed dataset which is then combined using ensemble. The dataset used is first preprocessed using the hair removal algorithm convolutional neural network using EfficientNet B0 – B7 and ResNet-50-v2 will be trained using ISIC 2020 dataset and ISIC 2020 dataset processed with hair removal algorithm.The model is evaluated using test data from ISIC 2020 dataset on area under the receiver operating characteristic curve (ROC AUC). The model trained will then be combined using ensemble where the result of the model will be averaged to give a combined prediction. The result of the test shows that the model trained is capable to classify melanoma and non-melanoma images. It is also shown that by removing hair from the skin image reduces the accuracy of th e model. Using Ensembling on the different models trained into one meta-model also increases the accuracy of the prediction giving a high final accuracy of 93.108%.

Rendering Karakter 3D Virtual secara Real-Time menggunakan Metode Light Estimation pada Augmented Reality Berbasis Lokasi Kevin Kevin; Liliana Liliana; Kartika Gunadi
Jurnal Infra Vol 9, No 2 (2021)
Publisher : Jurnal Infra

Show Abstract | Download Original | Original Source | Check in Google Scholar

Augmented reality applications are already widely available on mobile devices, but most augmented reality applications assume that light source always comes from above the object and its direction is always downwards so that the shadow is always right under the object, therefore a method is needed to estimate light so that the direction of shadow produced is more realistic, but can still be run on mobile devices.To answer the problem, light estimation method is used in real-time rendering of AR applications on mobile devices so that the shadow direction from virtual objects rendering is parallel and in the same direction as the shadow direction of real objects in their environment, but still uses resources that can be used on mobile devices.Results in this study indicate that the direction of shadow produced by light estimation method in indoor environment is quite accurate (about 33°) and light enough to be used on mobile devices, because the difference in FPS and RAM usage is almost the same as the usage of application without the use of light estimation method, although there is an increase in CPU and battery usage, it's small enough to still work on a mobile device.

Title

Found 13 Documents
Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title Search

Found 13 Documents Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title

Found 13 Documents
Search