Hand Detection Using Zoomed Neural Networks
Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review
Author(s)
Related Research Unit(s)
Detail(s)
Original language | English |
---|---|
Title of host publication | Image Analysis and Processing - ICIAP 2019, 20th International Conference |
Editors | Elisa Ricci, Samuel Rota Bulò, Cees Snoek |
Place of Publication | Cham |
Publisher | Springer |
Pages | 114-124 |
ISBN (electronic) | 978-3-030-30645-8 |
ISBN (print) | 978-3-030-30644-1 |
Publication status | Published - Sept 2019 |
Publication series
Name | Lecture Notes in Computer Science |
---|---|
Volume | 11752 |
ISSN (Print) | 0302-9743 |
ISSN (electronic) | 1611-3349 |
Conference
Title | 20th International Conference on Image Analysis and Processing (ICIAP 2019) |
---|---|
Location | |
Place | Italy |
City | Trento |
Period | 9 - 13 September 2019 |
Link(s)
Abstract
The object detection problem has been widely focused due to the development of personal cameras allowing the general population to have access to high end cameras. This has resulted in cameras with various perspectives, one of which is the Egocentric Perspective, like the GoPro cameras. This new perspective opens the possibility of having hand detection as a special problem, due to the hands containing enough information to be detected and even for hand recognition and users’s activity recognition. However, due to the perspective being new the databases are scarce, and most of them focus on generic object detection rather than hand detection. In this paper we address hand detection and hand disambiguation which focuses on detecting left and right hands as different objects.
This paper addresses these challenges by using the information of a left hand being the mirror image of the right hand for the hand disambiguation, and we also train a Neural Network to focus on the hand over all the image and another Neural Network to focus on the bottom area of the image, increasing the resolution as the hands go out of image, which is a characteristic of the hands in the Egocentric Perspective. In addition, we propose three Neural Network architectures using the hand and increase resolution bottom image information, and we compare them with current object/hand detection approaches.
This paper addresses these challenges by using the information of a left hand being the mirror image of the right hand for the hand disambiguation, and we also train a Neural Network to focus on the hand over all the image and another Neural Network to focus on the bottom area of the image, increasing the resolution as the hands go out of image, which is a characteristic of the hands in the Egocentric Perspective. In addition, we propose three Neural Network architectures using the hand and increase resolution bottom image information, and we compare them with current object/hand detection approaches.
Research Area(s)
- Neural Network, Hand detection, Hand disambiguation
Citation Format(s)
Hand Detection Using Zoomed Neural Networks. / Cruz, Sergio R.; Chan, Antoni B.
Image Analysis and Processing - ICIAP 2019, 20th International Conference. ed. / Elisa Ricci; Samuel Rota Bulò; Cees Snoek. Cham: Springer, 2019. p. 114-124 (Lecture Notes in Computer Science; Vol. 11752).
Image Analysis and Processing - ICIAP 2019, 20th International Conference. ed. / Elisa Ricci; Samuel Rota Bulò; Cees Snoek. Cham: Springer, 2019. p. 114-124 (Lecture Notes in Computer Science; Vol. 11752).
Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review