Monitoring Threatened Irish Habitats Using Multi-Temporal Multi-Spectral Aerial Imagery and Convolutional Neural Networks

Abstract

The monitoring of threatened habitats is a key objective of European environmental policy. Due to the high cost of current field-based habitat mapping techniques there is a strong research interest in proposing solutions that reduce the cost of habitat monitoring through increasing their level of automation. Our work is motivated by the opportunities that recent advances in machine learning and Unmanned Aerial Vehicles (UAVs) offer to the habitat monitoring problem. In this paper, a deep learning based solution is proposed to classify four priority Irish habitats types present in the Maharees (Ireland) using UAV aerial imagery. The proposed method employs Convolutional Neural Networks (CNNs) to classify multi-temporal multi-spectral images of the study area corresponding to three different dates in 2020, obtaining an overall classification accuracy of 93%. A comparison of the proposed method with a multi-spectral 2D-CNN model demonstrates the advantage of including temporal information enabled by the proposed multi-temporal multi-spectral CNN model.

Preliminary Classification of Selected Farmland Habitats in Ireland Using Deep Neural Networks

“…Yasir et al [ 38 ] describe available datasets that can be used for various habitat mapping problems but it doesn’t contain any farmland datasets. Some of the recent works in this regard are habitat mapping of marine, land, and benthic zones [ 39 , 40 , 41 , 42 , 43 , 44 , 45 , 46 , 47 , 48 , 49 ].…”