Oil Palm Fruit Dataset on Plantations for Harvesting Estimation

This dataset is a data set of oil palm FFB images taken directly from trees in commercial oil palm plantations in Indonesia. The dataset focuses on categorizing oil palm fruits into five stages of fruit maturity: Unripe (early stage fruit), Underripe (transitional phase fruit), Ripe (ripe fruit ready to harvest), Flower (representing the flowering stage), and Abnormal (fruit with typical characteristics due to potential disease or irregularity). This dataset has been validated by oil palm experts in categorizing the level of fruit maturity. It has been pre-processed so that it can be used for developing applications for detecting the level of FFB maturity in the plantation and calculating the number of fruits on oil palm trees with various methods and approaches. Various interests can use this dataset for research and application development such as students, lecturers, researchers, mobile-based application developers, machine learning and deep learning engineers, data science engineers, oil palm post-harvest experts, and other oil palm researchers. This dataset is useful for application development, application testing, and model validation of mobile-based applications or applications embedded in robots. The dataset was meticulously compiled in Central Kalimantan Province, Indonesia, through video recordings at 30 frames per second within an oil palm plantation. We have collected a total of 440 videos, each with varying lengths ranging from 8 seconds to 1 minute and 31 seconds. These videos are captured in a resolution of 320x640 pixels, providing a portrait orientation. The datasets have been split into data training, validation, and testing using composition 70:20:10 with the total images being 10207 for training, 2896 for validation, and 1400 for testing.

搜集汇总

背景与挑战

背景概述

该数据集聚焦于油棕果实的成熟度分类，包含在印度尼西亚种植园直接拍摄的图像，将果实分为未熟、半熟、成熟、开花和异常五个阶段，并经过专家验证。数据集已预处理，适用于开发检测成熟度和计算果实数量的应用，支持机器学习研究，数据收集自视频录制，总图像数达14503张，按70:20:10比例分割为训练、验证和测试集。