Kentaro-Machida/Matthiola-incana-Leaf-Image-Classification

Matthiola incana Leaf Image Classification dataset

Matthiola incana Leaf Image Classification(MiLIC) dataset is intended to create a model for classifying images of Matthiola incana leaves into single-flowered leaf or double-flowered leaf.

Matthiola incana

Matthiola incana, a member of the family Brassicaceae, is an ornamental plant cultivated worldwide for its beautiful flowers. This species exhibits two flower forms: single and double flowers. The market for single-flowered plants is extremely limited and their commercial value is lower than that of double-flowered plants. Therefore, to efficiently produce high-value M. incana, farmers select double-flowered plants during the seedling stage, a few weeks after sowing, based on cotyledon characteristics. However, this identification is not simple and can be extremely difficult for inexperienced individuals.Using this dataset, you can create a classification model for Matthiola incana.

About the MiLIC dataset

It contains data on the following varieties of M. incana:

• Quartet_Apricot
• Quartet_Blue
• Quartet_Marine
• Quartet_Rose

Although there are two existence folders for the same variety, such as Quartet_Apricot and Quartet_Apricot2, it is guaranteed that the M. incana contained in these folders are different individuals. To prevent data leakage, we recommend using either Quartet_Apricot or Quartet_Apricot2 as training data and the other as test data, and similarly for the other three varieties.

Each folder is further divided by {date}_{single or double}. The purpose of this data is to perform binary classification between single and double. Each JPG image is named using the date of data collection and the individual number. Each individual in Quartet_Apricot, Quartet_Apricot2, Quartet_Blue, ..., Quartet_Rose2 is completely separate. For example, Quartet_Apricot, Quartet_Apricot2, and Quartet_Blue each contain a JPG image of individual number 1, but these are completely separate individuals. The directory structure of Quartet_Apricot is shown below as an example.

Quartet_Apricot/
├── KA230307_single/
│   ├── KA230307 (1).JPG
│   ├── KA230307 (100).JPG
│   └── ... 
├── KA230307_double/
│   └── (JPG files)
├── KA230308_single/
│   └── (JPG files)
├── KA230308_double/
│   └── (JPG files)
├── KA230309_single/
│   └── (JPG files)
└── KA230309_double/
    └── (JPG files)

This dataset splits images not only by label but also by date, so when using it, we recommend downloading the dataset using the git clone command.

git clone https://huggingface.co/datasets/Kentaro-Machida/Matthiola-incana-Leaf-Image-Classification

Data collection

MiLIC dataset was created by agricultural experts at Mizuho Agricultural High School in Tokyo, Japan, in collaboration with the Nakajo Laboratory at Tokyo University of Agriculture and Technology.

We planted M.incana seeds in trays and tracked whether each individual became a single-flower or a double-flower. Images were taken between 2 and 3 weeks after planting the seeds.

License

The Creative Commons Attribution 4.0 International License (CC BY 4.0) allows users to share, copy, adapt, and build upon a work, even for commercial purposes, as long as proper credit is given to the original creator.