BUSCADOR RECOLECTA

The PFuji-Size dataset [2] includes the 3D point clouds of 6 Fuji apple trees containing a total of 615 apples and an additional 25 apples scanned in laboratory conditions. Structure-from-motion and multi-view stereo techniques were used to generate the 3D point clouds of the captured scene. Apple locations and ground truth diameter annotations are provided for assessing fruit detection and size estimation algorithms.

We provide a deep-learning method to better estimate the size of partially occluded apples. The method is based on ORCNN (https://github.com/waiyulam/ORCNN) and sizecnn (https://git.wur.nl/blok012/sizecnn), which extended Mask R-CNN network to simultaneously perform modal and amodal instance segmentation. The amodal mask is used to estimate the fruit diameter in pixels, while the modal mask is used to measure in the depth map the distance between the detected fruit and the camera and calculate the fruit diameter in mm by applying the pinhole camera model.

Multitask Deep Neural Network for Fruit Detection and Regresion of Fruit Diameters in RGB-D images (based on Detectron2). This project is an extension of the MaskRCNN architecture that allows to compute the diameter of fruits along with performing instance segmentation. The baseline for this project has been the detectron2 implementation of the MaskRCNN (https://github.com/facebookresearch/detectron2).

Fruit detection and 3D location in photogrammetry-derived 3D point clouds. This project is a matlab implementation to project image detections (instance segmentation masks) onto 3D point clouds generated using structure-from-motion.

The agricultural LiDAR data to evaluate scan matching techniques (AgLiMatch dataset) is comprised of a set of Velodyne VLP-16 LiDAR captures and the corresponding GNSS-RTK tracks acquired in a Fuji apple orchard using an autonomous platform. This dataset was used in [1] to evaluate scan matching techniques by comparing the platform path calculated using LiDAR scan matching techniques and the actual platform path ground truth measured with a GNSS-RTK system. The correspondence between each LiDAR file (inside /velodyne_data folder) and GNSS track file (inside /GNSS_data folder) is detailed in “Velodyne-GNSS_correspondence-data.xlsx” file. The relative position between the LiDAR sensor and the GNSS rover is shown in “experimental_setup.png”. Distance units are in mm.

Apple size estimation using photogrammetry-derived 3D point clouds. 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 project is a matlab implementation for fruit detection in 3D point clouds acquired with LiDAR sensor Velodyne VLP-16 (Velodyne LIDAR Inc., San Jose, CA, USA). This implementation was used to evaluate the LFuji-air dataset, which contains 3D LiDAR data of 11Fuji apple trees with the corresponding fruit position annotations.

Matlab implementation to evaluate RGB-D sensor performance in orchard environments. This project is a matlab implementation to evaluate RGB-D sensor performances by analysing RGB-D data acquired in different orchard conditions. The code follows the assessment methodology presented in [1], and it was used to evaluate the performance of Microsoft Kinect v2 by using the KEvOr dataset.

AKFruitData. This repository contents source code of two applications presented in the article https://doi.org/10.1016/j.softx.2022.101231 at the time of its publication in SoftwareX Journal. The emergence of low-cost 3D sensors, and particularly RGB-D cameras, together with recent advances in artificial intelligence, is currently driving the development of in-field methods for fruit detection, size measurement and yield estimation. However, as the performance of these methods depends on the availability of quality fruit datasets, the development of ad-hoc software to use RGB-D cameras in agricultural environments is essential. The AKFruitData software introduced in this work aims to facilitate use of the Azure Kinect RGB-D camera for testing in field trials. This software presents a dual structure that addresses both the data acquisition and the data creation stages. The acquisition software (AK_ACQS) allows different sensors to be activated simultaneously in addition to the Azure Kinect. Then, the extraction software (AK_FRAEX) allows videos generated with the Azure Kinect camera to be processed to create the datasets, making available colour, depth, IR and point cloud metadata. AKFruitData has been used by the authors to acquire and extract data from apple fruit trees for subsequent fruit yield estimation. Moreover, this software can also be applied to many other areas in the framework of precision agriculture, thus making it a very useful tool for all researchers working in fruit growing.

AKFruitData - ak_acquisition_system is a software solution for data acquisition in fruit orchards using a sensor system boarded on a terrestrial vehicle. It allows the coordination of computers and sensors through the sending of remote commands via a GUI. https://doi.org/10.1016/j.softx.2022.101231. AK_ACQS is a software solution for data acquisition in fruit orchards using a sensor system boarded on a terrestrial vehicle. It allows the coordination of computers and sensors through the sending of remote commands via a GUI. At the same time, it adds an abstraction layer on library stack of each sensor, facilitating its integration. This software solution is supported by a local area network (LAN), which connects computers and sensors from different manufacturers ( cameras of different technologies, GNSS receiver) for in-field fruit yield testing.