-79%
, ,

Artificial neural networks to passive acoustical identification of animals


In this project, MATLAB codes have been developed by MATLAB4Engineers team. Algorithms such as Extreme Learning Machine (ELM), Multi-Layer Perceptron (MLP), and different strategies for features extraction has been implemented and tested in MATLAB.

249 1.200

The development of a computer-based system capable of identifying animals automatically
from sounds they generate is described. The system uses time domain signal coding techniques
and artificial neural networks for discriminating between the sounds of different animal species
within a group. Results for British species of Orthoptera (bush crickets and crickets) show that it
is capable of discriminating 13 species with 100% success and zero misidentification under low
noise conditions. Results are also given for noise tests using 25 species of Orthoptera (including
grasshoppers). The approach can be applied to other animal groups such as birds, mammals and
amphibia; preliminary results are also presented for tests with 10 species of Japanese bird. The
approach is generic and has application in many fields including non-destructive testing and
physiological signal analysis.

The paper: download

For more details, watch videos in the following link: open videos

Based on 0 reviews

0.0 overall
0
0
0
0
0

Be the first to review “Artificial neural networks to passive acoustical identification of animals”

There are no reviews yet.