ArtAI

Welcome to ArtAI.

ArtAI is an artificial intelligence research institute established by Professor Yoon-Joong Kim to meet the needs of industrial sites and lead a beautiful and happy human life by deepening the theory of artificial intelligence technology that Professor Kim has researched during his tenure at Hanbat University.

The main field of interest is research on implementing technology to analyze and apply audio and video signals using HMM and deep learning models. For example, 1) Signal analysis, abnormal sound detection, recognition, and synthesis of time series data such as voice, machine sound, and bird sound. 2) We are conducting a research on situational awareness technology that detects people's faces, safety protective equipment, and license plates from video data with the characteristics of continuous two-dimensional images and judges the safety situation of workers.

Educational Materials

This is the enudational materials taught at Hanbat National University. The concepts and application methods corresponding to the basics of deep learning are summarized in six parts. Part 1: Overview of Machine Learning. Part 2: Supervised Learning, K-NN, Linear Regression, Ridge, Rasso regression model. Part 3: Logistic Regression, Binary Classification, Softmax Classification, Part 4: Application Development Tips such as Learning Rate, Data Preprocessing, Overfitting, Optimizer, batch Normalization, Learning Rate and NIST digit classifier example. Part 5: CNN, MNIST Digit Recognition Example. Part 6: RNN concept, Sine Wave Prediction,Text-Based Language Model (many-to-one), Text-based Language Model (many-to-many),Stock Price Prediction Model

Research Topics

We have achieved results in research that interprets audio, image and video data using artificial intelligence methods. For time-series data such as audio, speech recognition, synthesis, and voice emotion recognition were conducted. In a recent study, for video data frames, faces and license plates were detected using YOLOv4 models, and faces were recognized using ArcFace models and tracking techniques. Based on artificial intelligence technologies such as object detection, face recognition, and object tracking technologies described above, I have developed AISDS, a solution for de-identifying faces and license plates in video as a developer at Netvision Telecom Inc.