2024, Vol.10, No.5

Journal of Internet of Things and Convergence 2023 KCI Impact Factor : 0.89

pISSN : 2466-0078
https://journal.kci.go.kr/kiots

This paper describes a hybrid gesture interface based on MediaPipe that recognizes facial gestures and hand gestures using the MediaPipe Hands model and MediaPipe Face Mesh model, and then combines them. First, the presence of hands and faces is determined by individually detecting 3D hand landmarks of MediaPipe Hands model and 3D face landmarks of MediaPipe Face Mesh model from camera input frames, and then the face cursor position and face gestures, as well as the hand cursor position and hand gestures are recognized. Then, these are mixed in a user-friendly way to implement a user interface based on hand-face hybrid gestures. The proposed hand-face hybrid gesture interface based on MediaPipe has the advantage that the gesture mode is set to either the hand or the face, but the interface can be controlled freely using the face and hands without additional gestures for mode switching. In addition, the practicality and usefulness of the proposed hand-face hybrid gesture interface were confirmed through software operation experiments in Windows environment.

The application of machine learning models for trend prediction in the domestic stock market is increasing. In particular, utilizing machine learning is essential for analyzing and predicting complex time-series data, such as stock price data. This study proposes a machine learning system for financial time-series trend prediction, utilizing cloud computing services. First, for data collection, the serverless service of Amazon Web Services was employed, and the thresholds of technical analysis indicators were optimized through a genetic algorithm. The optimized indicators were then used as training data for Echo State Network, Recurrent Neural Network (RNN), and various machine learning classification models to predict the trend of each stock. Based on the predicted trends, backtesting was conducted, and the results showed that the average returns were 334% for ESN, 175% for RNN, and 199% for classification models. Therefore, this study suggests that machine learning exhibits high predictive power in domestic stock investment and holds various potential applications.

This study presents and verifies an easy, fast, and relatively cost-effective method for 6D pose estimation using industrial robots for bin picking in the manufacturing sector. Specifically, it details a method involving the integration of industrial robots with 2D cameras to ① acquire multi-view images of objects and collect training data, ② select variables from the collected data and implement a linear regression model, and ③ apply the trained model to estimate, verify, and evaluate the 6D pose of objects on industrial robots. The proposed data collection method and implemented linear regression model demonstrated statistically significant results. The estimated 6D poses were validated against ground true values and evaluated in their application to industrial robots, confirming their validity. By using feature point information extracted from images instead of direct image inputs as inputs to the regression model, the data size was reduced, enabling direct embedding on the robot. This research approaches the problem of spatial coordinates in 3D from a data analysis perspective, rather than from geometrical or computer vision perspectives.