Energy efficient object detection and measurement for smart glasses
We design and implement a novel object detection and measurement system called Lockon for smart glasses. Lockon takes the advantages of the mounting position of the smart glasses, and provides users two useful functions that can benefit a wide range of applications. Lockon can accurately locate the object of interest (OoI) in the view of the user and inform the user the position of the OoI in real-time, using the front-facing camera equipped on the smart glasses and advanced computer vision and image processing techniques. To conserve energy, Lockon implements a motion trigger to intelligently activate the object detection process only when it is necessary. Lockon can also accurately measure the dimension of the object, with a 3D ranging technique. This capability allows user to remotely estimate the dimension of the object. We implement Lockon on Google Glass Explorer Edition, and evaluate the performance of Lockon using extensive experiments. Our results indicate that Lockon can achieve high detection accuracy (0.95 true positive rate and $4.3 \times 10^{-7}$ false positive rate), low object dimension measurement error (3.3\% when distance is less than 1 m), and low delay (300 ms).
Read
- In Collections
-
Electronic Theses & Dissertations
- Copyright Status
- In Copyright
- Material Type
-
Theses
- Authors
-
Yang, Jing
- Thesis Advisors
-
Xing, Guoliang
- Committee Members
-
Enbody, Richard J.
Liu, Xiaoming
- Date Published
-
2014
- Program of Study
-
Computer Science - Master of Science
- Degree Level
-
Masters
- Language
-
English
- Pages
- vi, 41 pages
- ISBN
-
9781303879784
1303879786
- Permalink
- https://doi.org/doi:10.25335/h13c-1h96