초록

용어보기논문에서 용어와 풀이말을 자동 추출한 결과로, 시범 서비스 중입니다. AI-Helper

Abstract ▼ AI-Helper

This paper proposes a 2-layer gesture spotting and recognition model using data obtained from a kinect sensor and applies the model to the shooting game application. The layer1 models the gesture and non-gesture patterns using HMM and detects the start and end points of gesture using an adaptive thr...

This paper proposes a 2-layer gesture spotting and recognition model using data obtained from a kinect sensor and applies the model to the shooting game application. The layer1 models the gesture and non-gesture patterns using HMM and detects the start and end points of gesture using an adaptive threshold model based on HMM. Given a gesture segment from the layer1, the layer2 recognizes the gesture using CRF model. In particular, we propose the majority voting method for probabilities of accumulative sequences to reduce the errors and improve the recognition performance. To evaluate the performance of the proposed method, we define the arm gestures used in shooting game application and collect the dataset from a kinect sensor. The proposed method obtains the 97.54% spotting reliability and 100% recognition rate. We also compare the recognition rate between proposed model and HMM/CRF model to show the effectiveness of the proposed method.

주제어