Close

Initial to-do list

A project log for Automated Elephant-detection system

Real-time machine-vision based detection of elephants in rural areas: providing warnings to locals and thus prevent human-elephant conflict

neil-k-sheridanNeil K. Sheridan 03/19/2016 at 22:110 Comments

To-do List

1. The most challenging part of the project will be feature extraction from the obtained elephant images and similarity comparison between these and known elephant images from a database. I propose to feature-extract using the Haar wavelet technique (e.g. http://www.mathworks.com/matlabcentral/fileexchange/29049-wavelet-decomposition-for-images )

First will be testing of feature-extraction and comparison techniques using images of cardboard elephant shapes (front view, side view, adult/juvenile elephants) with a database of cardboard elephant shapes!

2. Design of prototype hardware elephant-detection device: comprising raspberry Pi, camera module [should be fairly easy]

3. Testing of prototype hardware and software with cardboard elephant shapes [will the Raspberry Pi be able to do this?]

4. Feature extraction from images of real elephants and comparison with known images of real elephants using prototype elephant-detection device hardware at a local safari park

Discussions