Open AccessExperimental evaluation of Data Augmentation heuristics for plant identification systems based on Deep Learning
Author(s) -
Luciano Araújo Dourado Filho,
Rodrigo Tripodi Calumby
Publication year - 2021
Language(s) - English
Resource type - Conference proceedings
DOI - 10.5753/sbiagro.2021.18384
Subject(s) - computer science , convolutional neural network , heuristics , artificial intelligence , machine learning , identification (biology) , generalization , deep learning , artificial neural network , scope (computer science) , data modeling , training set , plant identification , data mining , database , mathematical analysis , botany , mathematics , biology , programming language , operating system
Data augmentation (DA) allows increasing datasets for training machine learning models that demands large amounts of data. In real-world applications in which data may not be abundant enough and data acquisition is not easy, DA enables increasing diversity and introducing model generalization. In this work we evaluate several DA techniques and combining approaches to extend image datasets used to train plant species recognition models. We experimentally validated Deep Convolutional Neural Networks (DCNN) with several datasets obtained from common augmentation techniques and combinations. The results allowed the identification of the Translate + Crop augmentation policy as the most effective within the scope of evaluation.