Publications
SARLBP: Scale Adaptive Robust Local Binary Patterns for Texture Representation(Submitted for publication in IEEE Transactions on Image Processing (IEEE TIP))
This paper develops a novel texture feature descriptor called scale adaptive robust local binary pattern that utilizes both micro- and macro- level texture information by introducing adaptive scale selection and a novel encoding scheme.
STMO-GA: Sequential Transfer with Multi-Objective Based Genetic Algorithm for Feature Selection of High-Dimensional Problems (Accepted for publication in IEEE World Congress on Computational Intelligence (IEEE WCCI 2024). Yokohama, Japan.)
This paper introduces a method that combines sequential transfer learning with a multi-objective genetic algorithm to perform feature selection on datasets with an extreme case scenario where the number of instances is significantly limited compared to the number of dimensions.
Crop Residue Cover Percentage Estimation from RGB Images Using Transfer Learning and Ensemble Ordinal Regression (Accepted for publication in Journal of American Society of Agricultural and Biological Engineers (ASABE))
This paper utilizes transfer learning technique with an ensemble SVM ordinal regression model strategy facilitated estimating residue based on probabilistic estimates of the expert classifiers. This research confirms the utility of high-resolution RGB imagery to quantify residue cover in agricultural fields.
Classification of Crop Residue Cover in High-Resolution RGB Images using Machine Learning (Published in Journal of American Society of Agricultural and Biological Engineers (ASABE))
This paper develops a machine learning framework to estimate residue cover in RGB images taken at three resolutions. This research confirms that an automated system developed using machine learning is a viable strategy to estimate residue cover from RGB images obtained by hand-held or UAV platforms.
Classifying Cover Crop Residue from RGB Images: A Simple SVM Versus A SVM Ensemble (Published in IEEE Symposium Series on Computational Intelligence (SSCI). Orlando, Florida, USA)
This paper develops a SVM ensemble method with a hierarchical structure, to estimate residue cover in RGB images. This study represented a gain with respect to accuracy, compared to simple SVM classifier.
Capability of High-Resolution RGB Imagery to Accurately Document Residue in Row-Crop Fields (Published in Journal of Soil and Water Conservation (JSWC))
This paper documents precision error in field tape readings that results in lower accuracy in readings than in expected with a binomial distribution, and proposes bullseye grid approach which is an image-based method to improve the accuracy of residue estimates by facilitating method less prone to reader bias.