Automatic Identification of Tree Species From Sentinel-2A Images Using Band Combinations and Deep Learning

• Published in: IEEE geoscience and remote sensing letters Vol. 21; pp. 1 - 5
• Main Authors: Vaghela Himali, P., Raja, R. A. Alagu
• Format: Journal Article
• Language: English
• Published: Piscataway The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2024
• Subjects: Algorithms; Deep learning; Feature extraction; Image analysis; Image processing; Machine learning; Neural networks; Plant species; Remote sensing; Satellite imagery; Species identification; Trees
• ISSN: 1545-598X; 1558-0571
• DOI: 10.1109/LGRS.2024.3354814

Tree species identification using satellite images has been a prominent research topic in the field of remote sensing image analysis. So, this letter discussed an approach that explores various band combinations and deep learning models to identify tree species in the Madurai region from Sentinel-2A images. In the subject of identifying tree species, many machine-learning algorithms have been created. However, the ML model requires user intervention in selecting the features to process, which is time-consuming and based on trial and error. Owing to the existence of such gaps, this letter discusses a hybrid deep learning approach where feature extraction is performed using neural network blocks such as VGG, MobileNet, and ResNet, and classification is performed using a random forest (RF) classifier. Among all combinations, ResNet-RF gives 90.75% accuracy, and it outperforms gray level co-occurrence matrix (GLCM)-RF and other state-of-the-art deep learning models.