A Multi-Stage, Pixel-Level Annotated Apple Dataset for Precision Agriculture Research

The dataset of apples was collected from September 2017 to October 2019 in an experimental orchard at the College of Horticulture, Northwest A&F University (Yangling, Shaanxi, China). The dataset consists of 1,124 apple images captured under varying natural lighting and weather conditions, covering multiple growth stages—from immature green apples to mature red targets—including transitional color variations. The dataset is classified into three distinct categories: immature (green), semi-mature (transitional), and mature (red). Table 1 gives a detailed description and visualization for each categories. The dataset contains 1,406 annotated images capturing apples across three developmental stages: 532 images with immature (green) apples, 316 with semi-mature (color-transitioning) apples, and 384 with mature (red) apples. The non-additive total reflects natural orchard conditions where individual images frequently contain multiple maturity stages—either three stages simultaneously or two transitional stages coexisting. This indicates that asynchronous fruit development occurs within orchards. Across all images, we annotated 2,573 distinct apple instances: 921 immature (green), 813 semi-mature (green-red color transition), and 839 mature (red) apples.