Papers by Elizabeth Irenne Yuwono
Applied Soft Computing, Nov 1, 2022
Information Technology & People, 2021
PurposeThe paper proposes a privacy-preserving artificial intelligence-enabled video surveillance... more PurposeThe paper proposes a privacy-preserving artificial intelligence-enabled video surveillance technology to monitor social distancing in public spaces.Design/methodology/approachThe paper proposes a new Responsible Artificial Intelligence Implementation Framework to guide the proposed solution's design and development. It defines responsible artificial intelligence criteria that the solution needs to meet and provides checklists to enforce the criteria throughout the process. To preserve data privacy, the proposed system incorporates a federated learning approach to allow computation performed on edge devices to limit sensitive and identifiable data movement and eliminate the dependency of cloud computing at a central server.FindingsThe proposed system is evaluated through a case study of monitoring social distancing at an airport. The results discuss how the system can fully address the case study's requirements in terms of its reliability, its usefulness when deployed ...
Automatic Speech Recognition (ASR) atau speech to text adalah bidang teknologi identifikasi ucapa... more Automatic Speech Recognition (ASR) atau speech to text adalah bidang teknologi identifikasi ucapan manusia dalam bentuk teks transkripsi. Penelitian ini merupakan studi pada format masukan dan luaran speech to text, yaitu audio (ucapan) dan teks. Studi berfokus pada karakteristik dan format sinyal audio, pemrosesan sinyal audio secara digital dan relasinya dengan modul speech to text, pengetahuan linguistik, karakteristik dan format teks, serta isu pengembangan modul speech to text. Sinyal audio untuk ucapan memiliki beberapa karakteristik unik yang membedakannya dengan sinyal audio lain. Karakter-karakter ini merupakan fitur yang digunakan untuk identifikasi ucapan dalam sinyal audio masukan. Dalam modul speech to text sinyal digital mengalami beberapa proses sebelum identifikasi ucapan dilakukan. Proses sinyal digital ini dilakukan untuk memperoleh sinyal ucapan dengan tingkat kebisingan terendah dan hasil akurasi tinggi. Beberapa proses tersebut antara lain: sampling, kuantisasi,...
Virtual reality has been implemented in many fields recently escpecially in education because its... more Virtual reality has been implemented in many fields recently escpecially in education because its capability to produce a virtual world and take users to experience in different level with lower cost. The users will interact with the virtual world using their body or some parts of body such us head, hand, or voice. The problem of recognition accuracy level is still a challenging problem. This research is focused on comparing head movement recognition algorithms in a simple educative mobile application. Accelerometer sensor and RGB camera in Kinect are used to capture five basic head movementsl nodding, shaking, looking up, looking down, tilting. Three different algorithms are used to recognize the movement; backpropagation neural network, dynamic time wrapping and convolutional neural network. The result reveals that accelerometer-based dynamic time wrapping method is the best method in recognizing the head movement with 80o/, accuracy level, followed by backpropagation neural netwo...
Based on Internet Live Stats data, more than 2 billions information has been accessed daily of in... more Based on Internet Live Stats data, more than 2 billions information has been accessed daily of internet user daily. Amatriain said that trend of searching method has been deprecated. Trend of recommendation method has taken the position nowadays. Information must been processed to be transformed into recommendation and recommendation will reveal hidden information to the right position. One of recommendation method is people-to-people recommendation because one of the most accessed media in internet is social media. This research will discuss about comparison between hybrid algorithm in people-to-people recommendation. There are three algorithms which will be compared: hybrid content-collaborative reciprocal (without using weight) algorithm, hybrid contentcollaborative reciprocal (using weight) algorithm, and interactive-based + decision tree algorithm. These algorithm will be implemented in recommending workout partner using “FitParners”, Android-based mobile application and used 2...
2019 16th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS), 2019
Deep learning-based person re-identification faces a scalability challenge when the target domain... more Deep learning-based person re-identification faces a scalability challenge when the target domain requires continuous learning. Service environments, such as airports, need to recognize new visitors and add new cameras over time. Training-at-once is not enough to make the model robust to new tasks and domain variations. A well-known approach is fine-tuning, which suffers forgetting problem on old tasks when learning new tasks. Joint-training can alleviate the problem but requires old datasets, which is unobtainable in some cases. Recently, Learning without forgetting (LwF) shows its ability to mitigate the problem without old datasets. This paper extends the benefit of LwF from image classification to person re-identification with further challenges. Comprehensive experiments are based on Market1501 and DukeMTMC4ReID to evaluate and benchmark LwF to other approaches. The results confirm that LwF outperforms fine-tuning in preserving old knowledge and joint-training in faster training.
Lecture Notes in Electrical Engineering, 2016
As the media of communication for people with hearing and speech disabilities, the importance to ... more As the media of communication for people with hearing and speech disabilities, the importance to bridge the communication gap between them and normal people using sign language has become significance. This research proposed a model for the development of sign language recognition technology using Microsoft Kinect and convolutional neural network (CNNs). The proposed model succeeds in recognizing 10 dynamic Indonesian sign language words on complex background. There are total of 100 gesture image sequences containing color and depth data, perform by different users. The classifier consists of two CNNs and one ANN. The first CNN is to extract hand feature from color data, while the other is to extract hand feature from depth data. The training consists of three modes by applying drop-out and data augmentation and achieves the highest validation rate on 81.60 % and test result on 73.00 %.
Uploads
Papers by Elizabeth Irenne Yuwono