Introduction
We have collected and recorded multiple image datasets of strawberries, comprising the StrawDI dataset obtained from official websites, strawberry images captured from plantations, images of strawberries grown and photographed under laboratory conditions, as well as a strawberry image dataset retrieved from Google Dataset Search.
In addition, in order to further enhance the comprehensiveness of our digital image dataset of strawberries, we employed data augmentation techniques on Roboflow to expand the original collection of 5000+ strawberry images to a total of more than 17,200.
Collaborators
- Yongming CHEN*
- Xiaoxuan WANG*
- Feiyu ZHAO*
- Feiyang YU*
- Huan LIU*
*Intelligent Robotics Laboratory, School of Electrical Engineering and Automation, Hubei Normal University.
Datasets
We have created a comprehensive and large-scale digital image dataset for strawberry object detection. It has several features:
- Over 17,200 pictures
- 3 object categories
- 3100 pictures from StrawDb
- 1046 images from Google Dataset Search
In addition to strawberry datasets, our laboratory also has nearly 10000 pictures of campus masks dataset and a large aircraft data set.
Examples
Here are some examples of the Strawberry Digital Image Dataset we have used, with the original images on the left and the images using data augmentation techniques on the right.
| Original image | With data augmentation techniques |
|---|---|
 |
 |
 |
 |
Intelligent Robot Laboratory
Basic information of members
| Name | Status | Research direction | |
|
Ph.D supervisor |
1.Intelligent robot system 2.Deep learning and computer vision 3.Edge computing and cloud computing 4.Application of Transformer in computer vision tasks |
|
|
|
Ms. Xiaoxuan WANG |
M.E candidate |
|
|
|
Mr. Feiyu ZHAO |
M.E candidate |
|
|
|
Mr. Feiyang YU |
M.S candidate |
|
|
|
Mr. Huan LIU |
M.E candidate |
|
Photos of the members
 |
 |
 |
|---|---|---|
 |
 |
*From top to bottom, from left to right, the corresponding names are Yongming CHEN, Xiaoxuan WANG, Feiyu ZHAO, Feiyang YU and Huan LIU.
Publication of scientific research papers
Overview
Images
The datasets contain rich strawberry data in various scenarios, including traditional field cultivation and small-scale cultivation in pots, as well as elevated stereo cultivation methods, providing the model with a wide range of scenarios and situations during training, improving the robustness of the model.In addition, using roboflow for data augmentation can increase the quantity and diversity of training datasets, thereby enhancing the model's generalization performance, prediction accuracy, and robustness.
Annotations
According to the physiological information such as color and growth stage of strawberries, they can be roughly divided into three stages: raw, turning, and ripe. In the raw stage, strawberries are in the period of green ripening to white ripening, the fruit continues to grow, and the color changes from green to white. In the turning stage, strawberries enter the stage of color change, and the surface shows pink or light red. In the ripe stage, strawberries are fully matured, with more than 80% of the skin presenting bright red color, the fruit softening, and the sugar and nutrient content reaching the highest point.
Specific information
| Origin of datasets images | Number of images before data augmentation | Annotation information of images | Data augmentation information of images | Number of images after data augmentation |
|---|---|---|---|---|
|
train (2800 images) validation (100 images) test (200 images) |
raw turning ripe |
Flip: Horizontal;Crop: 0% Minimum Zoom, 15% Maximum Zoom;Rotation: Between -15° and +15°;Brightness: Between -25% and +25% |
StrawDI_Db1_DA: train (6448 images) validation (1612 images) |
|
|
Google Dataset Search(StrawDI_Db2) |
train (837 images) validation (209 images) |
Flip: Horizontal;Crop: 0% Minimum Zoom, 15% Maximum Zoom;Rotation: Between -15° and +15°;Brightness: Between -25% and +25% |
StrawDI_Db2_DA: train (3426 images) validation (830 images) |
|
|
small-scale cultivation in pots(StrawDI_Db3) |
592 images |
|||
|
elevated stereo cultivation in strawberry plantations(StrawDI_Db4) |
1042 images |
Flip: Horizontal;Crop: 0% Minimum Zoom, 15% Maximum Zoom;Rotation: Between -15° and +15°;Brightness: Between -25% and +25% |
StrawDI_Db3_DA: train (3986 images) validation (997 images) |
|
|
a total of 5780 images |
a total of 17,299 images |
COCO Explorer
COCO 2017 train/val browser (123,287 images, 886,284 instances). Crowd labels not shown.
Original images
Images using data augmentation techniques
License
These datasets are made freely available to academic and non-academic entities for non-commercial purposes such as academic research, teaching, scientific publications, or personal experimentation. Permission is granted to use the data given that you agree.
1. Overview
The strawberry object detection task refers to the automatic identification and localization of strawberries in images or videos, and the bounding of a rectangular box for each identified strawberry in the image. This task is typically implemented using computer vision algorithms and technologies, such as object detection models in deep learning (such as Faster R-CNN, YOLOv5, etc.). The applications of this task include strawberry agricultural production and trade management.
2. Strawberry Maturity Detection System
Introduction
A strawberry maturity detection system is a system that utilizes computer vision technology to detect the maturity of strawberries. The principle of the maturity detection method used by the system is to capture strawberry images through a camera, process the images, extract the color features of the strawberries, and then determine the maturity of the strawberries based on these color features. The application of a strawberry maturity detection system can help strawberry farmers and pickers improve their harvesting efficiency, reduce losses, and also improve product quality and market competitiveness.
Features
- Model select
- Input selection (local file, camera, RTSP video stream)
- IoU/Confidence/Latency adjustment
- Play/Pause/Finished
- Statistical test results
- After the test is completed, the test results are automatically saved
Real-time video detection
The significance of creating a comprehensive digital image dataset of strawberries includes the following aspects:
1.Sharing data resources: By creating a strawberry dataset website, strawberry-related data resources can be consolidated on one platform, making it easier for users who need to access, download, and share data, promoting data resource sharing and collaborative research.
2.Algorithm and model training and testing: The strawberry digital image dataset can serve as a basic data for algorithm and model training and testing, helping researchers adjust and optimize models to improve accuracy and efficiency.
3.Promoting research cooperation: The strawberry dataset website can become a platform for collaboration among strawberry researchers, facilitating joint completion of research projects, exchange of help and knowledge sharing.
4.Promoting industrial development: The strawberry dataset website can promote the development of the strawberry industry, improving the efficiency and level of strawberry production, quality, and sales through data analysis and mining, and promoting the sustainable development of the strawberry industry.