File(s) under permanent embargo
Automatic detection of moving baw baw frogs in camera trap videos
conference contribution
posted on 2018-10-05, 00:00 authored by Gelayol Golkarnarenji, Abbas KouzaniAbbas Kouzani, Nathan Semianiw, D Goodall, D Gilbert, Don DriscollDon DriscollThis paper presents the design and implementation of a motion detection algorithm for processing of video sequences captured by a purpose-built camera trap, removing of unwanted frames, and reducing of the size of the video sequences. The camera trap is used to create a dataset containing H.264 video sequences of the critically endangered Baw Baw Frog Philoria frosti. A motion detection method based on the optical flow Farneback algorithm is then developed to reduce the size of the video sequences by detecting the frames that contain moving frogs, keeping and storing the detected frames, and discarding the frames that do not contain moving frogs. Twenty video sequences containing a total of 13770 frames are used in this study. The frames with moving frogs are detected with sensitivity of 0.993, specificity of 0.992, and accuracy of 0.992. The original video sequences are then replaced by the newly created video sequences that are much smaller in size with an average reduction of 80.4% for the test video sequences used in this study.
History
Event
Mechatronics and Automation. Conference (2018 : Changchun, China)Pagination
1112 - 1116Publisher
IEEELocation
Changchun, ChinaPlace of publication
Piscatawy, N.J.Publisher DOI
Start date
2018-08-05End date
2018-08-08ISBN-13
9781538660720Language
engPublication classification
E Conference publication; E1 Full written paper - refereedTitle of proceedings
ICMA 2018 : Proceedings of IEEE International Conference on Mechatronics and AutomationUsage metrics
Categories
No categories selectedKeywords
Licence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC