Thesis Topic Details

Topic ID:
546
Title:
Visual monitoring of wildlife by computer
Supervisor:
Andrew Taylor
Research Area:
Associated Staff
Assessor:
Malcolm Ryan
Topic Details
Status:
Active
Type:
R & D
Programs:
CS CE BIOM BINF SE
Group Suitable:
Yes
Industrial:
Pre-requisites:
Strong programming skills.
Description:
This is project is develop software to detect and classify animals in
a stream of images coming from cameras in the natural environment.

For example, one dataset is several weeks of video of echidna's burrow
entrance. The software should detect the echidna's entering and leaving
the burrow and hence provide a sumamry of its activity to biologists.

Another dataset will be video of an watering point in arid inland where
rare birds and mammals come to drink. The goal is develop solar-powered
system which will left at the watering points for a year or more and
can can capture and store images of every animal that visits day or night.

The overall project will involve:

a) detection of scene change in images e.g. the echidna leaving its burrow
Clouds, shadows and moving vegetation make this non-trivial - good
sub-problem for a thesis.

b) segmentation of images, e.g. determining which part of the image
contains the echidna. The time series of images makes this much more
tractable than it would be otherwise.

c) unsupervised clustering of images - finding all the images into groups,
e.g. one group might contain green-and-yellow birds. Challenging but
plenty of room for progress in a 4th year thesis.

d) supervised classification of images, e.g. finding all the images
containing kangaroos. Challenging but plenty of room for progress in
a 4th year thesis.

A single student's thesis will likely only tackle one or two of
the above.

Students can choose their approach to the problem. Techniques are likely
to come from one or more of the fields of

a) image processing
b) neural networks
c) machine learning
d) statistics
Comments:
--
Past Student Reports
 
No Reports Available. Contact the supervisor for more information.

Check out all available reports in the CSE Thesis Report Library.

NOTE: only current CSE students can login to view and select reports to download.