Thumbnail Image
Item

Accelerating Gesture Detection DNNs on Mobile Devices with Model Pruning and Whole Application Optimization Co-Design

Wang, Charles
Abstract
Recognizing hand gesture detection is becoming increasingly essential as humans and computers interact closer and closer; we propose a framework for near real-time 3-D hand gesture reconstruction on mobile devices. Specifically, we accomplished a whole-application optimization on the hand gesture recognition and rendering task by parallelism and pipelining. With optimization of the model itself, we can achieve a near-real-time performance on Android cell phones, with the core design idea applicable to other platforms. Moreover, we present extensive data on partition tasks to achieve efficient pipelining.
Description
Date
2022-05-01
Journal Title
Journal ISSN
Volume Title
Publisher
Collections
Undergraduate Honors Theses
Show all metadata
Download Dataset
Files
Charles_Wang_Honors_Thesis.pdf
Adobe PDF, 900.06 KB
Rights Holder
Usage License
Embargo
Research Projects
Organizational Units
Journal Issue
Keywords
Citation
Advisor
Ren, Bin
Li, Qun
Li, Chi-Kwong
Department
Computer Science
DOI
URI
https://scholarworks.wm.edu/handle/internal/11264
Embedded videos