In Field Fruit Sizing Using A Smart Phone Application.

Area of Science:

• Agricultural technology
• Computer vision
• Horticulture

Background:

• Accurate fruit sizing is crucial for crop health assessment and yield prediction.
• Mobile phones offer a readily available, sensor-rich platform for agricultural data collection.

Purpose of the Study:

• To develop and validate an Android application for in-field fruit size measurement using mobile phone cameras.
• To assess the accuracy and efficiency of the developed application across various fruit types.

Main Methods:

• Development of the
• FruitSize
• Android application utilizing phone camera imaging.
• Implementation of image processing and object segmentation techniques (OpenCV) for fruit isolation.
• Calculation of fruit dimensions using camera parameters and allometric relationships.
• Integration of geolocation data for spatial mapping of fruit measurements.

Main Results:

• The application achieved a typical assessment rate of 240 fruits per hour.
• Under controlled lighting, Root Mean Square Errors (RMSEs) for fruit diameter ranged from 2.0 mm (apple) to 3.8 mm (mandarin).
• For mangoes, in-field RMSEs for length and width were 5.5 mm and 4.6 mm, respectively, benchmarked against manual caliper measurements.

Conclusions:

• The
• FruitSize
• application provides a practical and efficient method for in-field fruit sizing.
• The technology demonstrates potential for improving crop monitoring and yield estimation in agriculture.
• Mobile phone-based imaging offers a viable alternative to traditional measurement tools for fruit assessment.