Outdoor oil palm fruit ripeness dataset

Area of Science:

• Agricultural Science
• Computer Vision
• Machine Learning

Background:

• Accurate oil palm fresh fruit bunch (FFB) grading is crucial for optimizing oil quality and cultivation practices.
• Existing methods for FFB ripeness assessment often lack comprehensive, diverse datasets, especially for outdoor, real-world conditions.
• Image processing and machine learning offer potential for automated and objective FFB grading.

Purpose of the Study:

• To introduce and describe a novel dataset of oil palm fresh fruit bunch (FFB) images.
• To facilitate research in automated FFB ripeness detection and classification using image processing.
• To support the development and validation of machine learning models for outdoor FFB grading.

Main Methods:

• Collected images of oil palm fresh fruit bunches (FFB) from Malaysian plantations.
• Captured FFB images from various angles, both on the tree and on the ground.
• Classified FFB images into five ripeness categories: damaged, empty, unripe, ripe, and overripe, with expert ground truth labeling.

Main Results:

• The dataset contains diverse FFB images with detailed ripeness classifications.
• It highlights color variations and external characteristics indicative of FFB ripeness.
• Provides a valuable resource for developing and testing machine learning algorithms for automated FFB grading.

Conclusions:

• The FFB image dataset is a significant resource for advancing automated oil palm fruit classification and grading.
• This dataset enables the development of robust machine learning pipelines for real-world outdoor FFB assessment.
• It has the potential to improve oil palm cultivation, increase yield, and optimize oil quality.