Multi-class-peach-RGB-D-dataset

Introduction

The Multi-class-peach-RGB-D-dataset is composed by 2627 multi-modal images of 577 naked peaches and 2050 bagging peaches on tree filelds captured using Microsoft Azure Kinect camera. Each image contains 3 different modalities, including RGB image, Depth image and Infrared image.

All images were aligned with the RGB image and manually labelled in 4 classes by occlusion status: non-occluded, occluded by leaves, occluded by branches and occluded by fruits. Find annotations in ".txt" format inside "annotations" folder. The following informations are some of the camera parameters.

Color camera:

|Resolution|1280 x 720|
|FOV       |90° x 59° |
|Format    |   png    |

Depth camera:

|Resolution|1024 x 1024|
|Mode      |    WFOV   |
|FoI       |120° x 120°|

Folder structure:

Multi-class-peach-RGB-D-dataset
├── depth
│    ├── 1.png
│    └── ....
├── rgb
│    ├── 1.png
│    └── ....
├── infrared
│    ├── 1.png
│    └── ....
├── annotations
     ├── 1.txt
     └── ....       

For more dataset, please contact Email:tsing999@stu.ahau.edu.cn, thanks.