2026-10-13T22:12:48Z https://www.scidb.cn/oai

10.57760/sciencedb.j00240.00237 2026-09-01T06:55:43Z

2026-09-01 MelonNightColor: a multi-color artificial illumination image dataset for nighttime greenhouse melon segmentation doi:10.57760/sciencedb.j00240.00237 en MelonNightColor is a multi-color artificial lighting image dataset for greenhouse night melon segmentation. This dataset contains 782 sets (5474 images) of annotated nighttime melon images and 1218 sets (8526 images) of unlabeled images. During the acquisition process, the camera position and scene were kept relatively stable, and only the artificial lighting RGB components were changed to form 7 real light source color images, including red, blue, green, cyan, purple, yellow, and white. The dataset provides pixel level semantic segmentation annotations, including three categories: melon fruit, melon stem, and background. The dataset is organized in two ways: single color and combination color, with a total of 11 sub datasets: 7 single color datasets—— data_red、data_blue、data_green、data_cyan、data_purple、data_yellow、data_white， Each contains 782 images; Four combined color datasets - data-red_blue, data-red_green, data_green-blue - each contain 1564 images, and data-red_green-blue contains 2346 images. All datasets are divided into training set, validation set, and testing set in a 7:2:1 ratio, and are divided by shooting location (group) to ensure that 7-color images in the same group do not cross sets and avoid scene level data leakage. This dataset can be used for research on semantic segmentation of agricultural images, controlled lighting imaging, color domain shift analysis, and visual perception of facility agriculture. melon; image segmentation; artificial illumination; nighttime greenhouse; color-domain shift Liu Pengdong Lin Hanran Wu Maonian Zheng Bo Zhu Shaojun Shi Xingren EMBARGO https://creativecommons.org/licenses/by-nc-nd/4.0/ dataset Science Data Bank