Optimizing Data Processing in Animal Sensing Systems using Deep Learning Algorithms for Efficient Habitat Monitoring

Document Type

Article

Publication Title

Journal of Wireless Mobile Networks Ubiquitous Computing and Dependable Applications

Abstract

Black Vue camera traps. Camera traps and GPS collars are some of the types of animal sensing technologies used for wildlife and habitat monitoring. However, the data collected by these systems can be voluminous, and how to mine that data for useful information has confounded researchers. Witnessing the ability of deep learning models (a subset of artificial intelligence (AI) models capable of learning from high-dimensional input) to analyze and derive biological insights from this massive dataset efficiently is stunning. They design a DL-based approach to compress data during the demanding process of scanning data in animal sensing frameworks, aiming to improve habitat surveillance. In the first phase, we will develop a data pipeline using deep learning to identify and classify animals in camera trap images. For this purpose, they propose deep convolutional neural networks, which have already achieved remarkable results in many classification tasks. That automation means researchers have more time and energy to analyze a greater amount of data. Here, we have part 1 next. Next, we will add a frontend and plug it into a system for handling data from multiple sensors, such as a camera trap, GPS collar, and weather sensor. The algorithm employs both deep and machine learning methods to analyze the information and infer knowledge about the animals' movement patterns and habitat selections.

First Page

298

Last Page

318

DOI

10.58346/JOWUA.2025.I3.017

Publication Date

9-1-2025

This document is currently not available here.

Share

COinS