2026-10-14T22:36:51Z https://www.scidb.cn/oai

10.57760/sciencedb.40726 2026-06-23T11:08:33Z

2026-06-23 CTSD-YOLO-Split: Citrus Time-Series Dataset for Object Detection and Segmentation (YOLO Format) & Test doi:10.57760/sciencedb.40726 en Citrus fruits go through four key phenological stages throughout their entire growth cycle: flowering (May), young fruit (June), green fruit (July October), and maturity (November December). There are significant differences in fruit size, color, surface texture, and differentiation from the background at different stages. To support the research on visual monitoring and yield estimation of citrus throughout the entire growth cycle, this study constructed the Citrus Time Series Dataset (CTSD), which covers the complete 8-month time series from flowering to maturity. The construction of this dataset aims to address the following key challenges: Obstruction problem: The main difficulty in detection is that fruits in the orchard are obstructed by branches and leaves, and fruits overlap with each other Scale changes dramatically: the fruit ranges from small flower clusters during flowering to large fruits with a diameter of 68mm during maturity, with a wide range of target scales Environmental changes: significant differences in lighting conditions, weather changes, and background complexity under different periods and cultivation modes  Citrus; Object Detection; Instance Segmentation; YOLO; Precision Agriculture; Time-Series Dataset Jiancheng Du EMBARGO https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank