Apple size estimation using photogrammetry-derived 3D point clouds

Introduction

This project is a matlab implementation for apple size estimation in 3D point clouds. Four different size estimation methods are implemented: largest segment, least squares, MSAC and template matching. This code was used in [1] to compare the performance of the mentioned methods by using the PFuji-Size dataset (not publicly available yet). Find more information in:

• In-field apple size estimation using photogrammetry-derived 3D point clouds: comparison of 4 different methods considering fruit occlusions [1] (submitted, not publicly available yet).

Preparation

First of all, create a new project folder:

mkdir new_project

Then, clone the code inside “new_project” folder:

cd new_project
git clone https://github.com/GRAP-UdL-AT/apple_size_estimation_in_3D_point_clouds.git

Prerequisites

• MATLAB R2020a (we have not tested it in other matlab versions)
• Computer Vision System Toolbox
• Statistics and Machine Learning Toolbox

Data Preparation

Inside the “new_project” folder, save the dataset folder “PFuji-Size_dataset” available at PFuji-Size dataset. (not publicly available yet. It will be made publicly available after the corresponding publication acceptance)

Launch the code

• Execute the file /new_project/apple_size_estimation_in_3D_point_clouds/test_apple_size_estimation_in_3D_point_clouds.m

Authorship

This project is contributed by GRAP-UdL-AT.

Please contact authors to report bugs @ [email protected]

Citation

If you find this implementation or the analysis conducted in our report helpful, please consider citing:

@article{Gené-Mola2020,
    Author = {{Gen{\'e}-Mola, Jordi and Sanz-Cortiella, Ricardo and Rosell-Polo, Joan R and Escol{\`a}, Alexandre and Gregorio, Eduard },
    Title = {In-field apple size estimation using photogrammetry-derived 3D point clouds: comparison of 4 different methods considering fruit occlusions},
    Journal = {Submitted},
    Year = {2020}
    doi = {https://doi.org/Submitted}
}