An Annotated Image Dataset for Apple Fruit Detection
Creators
- 1. School of Engineering and Management Vaud (HEIG-VD), HES-SO University of Applied Sciences and Arts Western Switzerland, Yverdon-les-Bains, Switzerland
- 2. JDC Electronic SA, Yverdon-les-Bains, Switzerland
Description
Dataset description
The dataset contains 60 images in JPG format, together with the corresponding annotation files.
The latter follow the YOLO annotation format and are stored as TXT files, where:
- '0' represents an apple fruit
- '1' represents the approximate outline of the central (dominant) tree in the image
Apple fruits are annotated using rectangular bounding boxes, while trees are annotated with custom polygonal shapes defined by multiple points.
The apple tree images were acquired in July 2024 from three different apple varieties (Gala, Golden, and Jazz) using:
- Samsung Galaxy Z Flip4 (resolution: 2252 × 4000 px, images denoted as Galaxy)
- Google Pixel 7a (resolution: 2268 × 4032 px, images denoted as MTP)
in an orchard located around Lake Geneva.
Nomenclature
The nomenclature for both JPG and TXT files follows the structure:
imageID-tree_side_variety_smartphone
where
- imageID is the unique identifier assigned to each image,
- tree denotes the specific tree depicted in the image,
- side indicates the orientation of the tree side on which the image was captured (N: north, S: south, E: east, O: west),
- variety specifies the apple variety (Gala, Golden, or Jazz), and
- smartphone identifies the device used to acquire the image, either the Samsung Galaxy (Galaxy) or the Google Pixel (MTP) smartphone.
Example: In 0d9d3d8f-REF_3_O_Golden_Galaxy.jpg, 0d9d3d8f corresponds to the imageID, REF_3 identifies the tree, O specifies that the image was captured from the west-facing side of the tree, which belongs to the Golden variety, and the image was acquired using a Samsung Galaxy Z Flip4 smartphone.
Files
annotation.zip
Files
(231.7 MB)
| Name | Size | Download all |
|---|---|---|
|
md5:a114ca33a582b41b35c41e5ab74bb56d
|
299.1 kB | Preview Download |
|
md5:fdbaf5efc7692b774e7d93af230ed0f3
|
231.4 MB | Preview Download |
Additional details
Funding
- HES-SO University of Applied Sciences and Arts Western Switzerland
- 130502/IA-RECHERCHE23-42