Strawberry Fruit Shape: Images and Morphometric Features

6,874 binary images of individual strawberry fruits, each standardized to 1000x1000 pixels, are paired with a CSV file containing 431 morphometric features. The dataset captures fruit from two distinct harvest dates, enabling temporal shape analysis. Features were extracted using methods including Elliptical Fourier Analysis, Generalized Procrustes Analysis, and EigenFruit Analysis.

Use Cases

• Training shape classification models based on 431 extracted geometric and Fourier descriptors.
• Analyzing temporal shape variation using images from two separate harvest dates.
• Benchmarking fruit segmentation algorithms on standardized 1000x1000px binary images.
• Studying the relationship between experimental plot design (plot ID) and fruit morphology.

Strengths

• Contains 6,874 consistently processed binary images, providing a substantial sample size for analysis.
• Offers a detailed feature set with 431 morphometric columns derived from multiple analytical methods.
• Includes raw RGB images of multiple fruits and processed individual images, allowing traceability.

Limitations

• Metadata completeness is low; specific column names, row counts for the CSV, and a precise last updated date are unavailable.
• The dataset's geographic origin and the exact time range of the harvests are not specified.
• Documentation references two different CSV filenames ('Fruit_image_data.csv' and 'Quantative_Strawberry_Features.csv'), creating ambiguity about the primary data file.

Provenance

Source

Mitchell J. Feldmann, University of California, Davis

Collection Method

Images were taken from field experiment plots on two harvest dates, followed by individual fruit extraction, binarization, and feature extraction.

Time Range

Covers two harvest dates, but specific years are not provided.