Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


Thesis_ShengDing_Finalver_mod.pdf (2.5 MB)

ETD Abstract Container

Abstract Header

A Detachable LSTM with Residual-Autoencoder Features Method for Motion Recognition in Video Sequences

Ding, Sheng
http://rave.ohiolink.edu/etdc/view?acc_num=osu160673417735023

Abstract Details

2020, Master of Science, Ohio State University, Electrical and Computer Engineering.
Motion recognition in video sequences is a challenging computer vision problem. Actions are represented as a series of frames in video environments, which can be easily understood by analyzing multiple frames' contents. In this thesis, we recognize human actions in a way similar to our observation of actions in real life, which is exploring the features of consecutive frames and the connection between them. Traditionally, feature extraction and recognition in video motion recognition are integrated, and the training time is lengthy [1-5]. Especially when new data is given, the time cost of retraining may be days which is too high, and the reliability for a new environment is low. We want to break down this process to reduce the difficulty of training, and at the same time, to find a reliable description of the process of feature extraction. In this thesis, we propose a detachable training motion recognition method by processing the video data using Residual Block Autoencoder (ResAE) and Long Short-Term Memory (LSTM) network. The proposed method can provide a reliable feature extraction and process long videos by analyzing the features in frame sequences. Experimental results show acceptable performance over 60% accuracy, which is promising, in action recognition using the proposed method on UCF-101 (action recognition dataset).
Xiaorui Wang (Advisor)
Wladimiro Villarroel (Committee Member)
Computer Engineering; Computer Science; Electrical Engineering

Recommended Citations

Citations

  • Ding, S. (2020). A Detachable LSTM with Residual-Autoencoder Features Method for Motion Recognition in Video Sequences [Master's thesis, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu160673417735023

    APA Style (7th edition)

  • Ding, Sheng. A Detachable LSTM with Residual-Autoencoder Features Method for Motion Recognition in Video Sequences. 2020. Ohio State University, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu160673417735023.

    MLA Style (8th edition)

  • Ding, Sheng. "A Detachable LSTM with Residual-Autoencoder Features Method for Motion Recognition in Video Sequences." Master's thesis, Ohio State University, 2020. http://rave.ohiolink.edu/etdc/view?acc_num=osu160673417735023

    Chicago Manual of Style (17th edition)

osu160673417735023
196
© 2020, some rights reserved.Creative Commons License A Detachable LSTM with Residual-Autoencoder Features Method for Motion Recognition in Video Sequences by Sheng Ding is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. Based on a work at etd.ohiolink.edu.
This open access ETD is published by The Ohio State University and OhioLINK.
Release 3.2.12