COMPUTER VISION AND MACHINE LEARNING FOR PLANT SENSING: APPLICATIONS FOR TRACKING IN AGRICULTURE

• Sponsoring Institution: Agricultural Research Service/USDA
• Project Status: ACTIVE
• Funding Source: USDA COOPERATIVE AGREEMENT
• Reporting Frequency: Annual
• Accession No.: 0429759
• Project Start Date: Sep 1, 2015
• Project End Date: Aug 31, 2020
• Program Code: [(N/A)]- (N/A)

Recipient Organization
MARQUETTE UNIV
(N/A)
MILWAUKEE,WI 53233

Performing Department
(N/A)

Non Technical Summary
(N/A)

Animal Health Component

85%

Research Effort Categories

Basic

15%

Applied

85%

Developmental

0%

Classification

Knowledge Area (KA)	Subject of Investigation (SOI)	Field of Science (FOS)	Percent
402	1110	2020	100%

Knowledge Area
402 - Engineering Systems and Equipment;

Subject Of Investigation
1110 - Apple;

Field Of Science
2020 - Engineering;

Keywords

computer

vision

tracking

insects

fruit

Goals / Objectives
The objective for this project is to explore the use of tracking algorithms and other techniques in computer vision for various applications, which may include tracking of insects for the monitoring of pest populations, or tracking of flowers and fruit for flower, or fruit load estimation.

Project Methods
Tracking algorithms are typically limited by needing to define the number of targets in a sequence beforehand. However, in agricultural applications, oftentimes, the number of objects is unknown. New tracking algorithms, such as the Random Finite Set method, allow for estimation not only of the position of the object, but also of the number of objects. With the relaxation of the requirement that the number of objects be known, these new tracking algorithms could then be applied to agricultural problems. Tracking is needed in many areas of agricultural automation, especially when robotics are involved. This project aims to investigate the use of tracking algorithms in an agricultural setting. Since it is unknown how well these algorithms will work in practice, this project may investigate other related topics where the aim is to localize objects of interest in images.