Development of Operation Estimation Method Based on Tracking Records Captured by Kinect

Autor:	Huaping Liu, Fuchun Sun, Guo Xuan, Guangyu Zhao, Sota Yoshii, Naoyuki Fujiwara, Bin Wang, Weihang Wu
Rok vydání:	2017
Předmět:	Training set Computer science business.industry Frame sequence Frame (networking) ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Process (computing) Tracking system Tracking (particle physics) Factory (object-oriented programming) Computer vision Artificial intelligence business Test data
Zdroj:	Communications in Computer and Information Science ISBN: 9789811052293 ICCSIP
DOI:	10.1007/978-981-10-5230-9_15
Popis:	In order to evaluate workers’ progress in the factory, we developed a model of operation estimation method based on tracking records captured by Kinect. In this paper, we use Kinect sensors to capture the human motion process (such as broadcast gymnastics), extract the skeleton frame sequence from training data as a template, then improve the DTW algorithm to match the skeleton frame in the test data and in the training data, to estimate the percentage of the action’s completion. The improved DTW algorithm achieves state-of-the-art performances on our dataset, higher than 90%.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::fbced974683217377519f59b4e479bb4 https://doi.org/10.1007/978-981-10-5230-9_15 Zobrazit plný text záznamu