Dataset Card for AGM Dataset

Dataset Summary

The AGM (AGricolaModerna) Dataset is a comprehensive collection of high-resolution RGB images capturing harvest-ready plants in a vertical farm setting. This dataset consists of 972,858 images, each with a resolution of 120x120 pixels, covering 18 different plant crops. In the context of this dataset, a crop refers to a plant species or a mix of plant species.

Supported Tasks

Image classification: plant phenotyping

Languages

The dataset primarily consists of image data and does not involve language content. Therefore, the primary language is English, but it is not relevant to the dataset's core content.

Dataset Structure

Data Instances

A typical data instance from the training set consists of the following:

{
  'image': <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=120x120 at 0x29CEAD71780>,
  'crop_type': 'by'
}

Data Fields

The dataset's data instances have the following fields:

• image: A PIL.Image.Image object representing the image.
• crop_type: An string representation of the crop type in the image

Data Splits

• Training Set:
  • Number of Examples: 972,858

Dataset Creation

Curation Rationale

The creation of the AGM Dataset was motivated by the need for a large and diverse dataset that captures various aspects of modern agriculture, including plant species diversity, stress detection, and crop health assessment.

Source Data

Initial Data Collection and Normalization

The images were captured using a high-resolution camera positioned above a moving table in an agricultural setting. The camera captured images of the entire table, which was filled with trays of harvested crops. The image capture process spanned from May 2022 to December 2022. The original images had a resolution of $1073{\times}650$ pixels. Each pixel in the images corresponds to a physical size of $0.5$ millimeters.

Annotations

Annotation Process

Agronomists and domain experts were involved in the annotation process. They annotated each image to identify the crops present and assign them to specific categories or species. This annotation process involved labeling each image with one of 18 distinct crop categories, which include individual plant species and mixtures of species.

Who Are the Annotators?

The annotators are agronomists employed by Agricola Moderna.

Personal and Sensitive Information

The dataset does not contain personal or sensitive information about individuals. It primarily consists of images of plants.

Considerations for Using the Data

Social Impact of Dataset

The AGM Dataset has potential social impact in modern agriculture and related domains. It can advance agriculture by aiding the development of innovative technologies for crop monitoring, disease detection, and yield prediction, fostering sustainable farming practices, contributing to food security and ensuring higher agricultural productivity and affordability. The dataset supports research for environmentally sustainable agriculture, optimizing resource use and reducing environmental impact.

Discussion of Biases and Known Limitations

The dataset primarily involves images from a single vertical farm setting therefore, while massive, includes relatively little variation in crop types. The dataset's contents and annotations may reflect regional agricultural practices and preferences. Business preferences also play a substantial role in determining the types of crops grown in vertical farms. These preferences, often influenced by market demand and profitability, can significantly differ from conventional open-air field agriculture. Therefore, the dataset may inherently reflect these business-driven crop choices, potentially affecting its representativeness of broader agricultural scenarios.

Additional Information

Dataset Curators

The dataset is curate by DeepPlants and AgricolaModerna. You can contact us for further informations at nico@deepplants.com etienne.david@agricolamoderna.com

Licensing Information

Citation Information

If you use the AGM dataset in your work, please consider citing the following publication:

@InProceedings{Sama_2023_ICCV,
    author    = {Sama, Nico and David, Etienne and Rossetti, Simone and Antona, Alessandro and Franchetti, Benjamin and Pirri, Fiora},
    title     = {A new Large Dataset and a Transfer Learning Methodology for Plant Phenotyping in Vertical Farms},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops},
    month     = {October},
    year      = {2023},
    pages     = {540-551}
}