Multispectral leaf-level analysis

The rapid detection of trees, which are suspected as infected by HLB, in large orchards, will enable growers to conduct PCR tests for such selected trees without delay, enabling fast decision making, reducing production losses, production costs, and even more important - prevention of HLB spreading.


Agrowing Ltd., an Israeli world leading provider of award-winning AI-Enabling multispectral sensors, and Vetorgeo, a Brazilian startup focused on Agriculture 4.0 solutions, collaborating with São Paulo University (UNESP) announce a significant breakthrough in leaf-level and aerial HLB detection in Citrus.

HLB is considered the main Citrus disease worldwide. The disease has no cure. It decreases oranges' production as infected trees must be removed from orchards as soon as they are detected to stop the disease’s spreading. Currently, the removal of infected trees is carried out after the visual confirmation of symptoms of HLB by trained inspectors teams. However the visual inspection takes long time, it is not sufficiently accurate, it is expensive and tedious.

VetorGEO’s research team has been developing a new detection method utilizing Agrowing’s high resolution 4 and 10 band multispectral sensors.

Preliminary tests were carried out in highlighted leaves using Agrowing’s 4 band 10MP per band sensor. The tests demonstrated the possibility to identify and isolate multispectral signatures of HLB symptoms in Citrus leaves.

The outstanding accuracy obtained in the preliminary test opened the possibility for the development of two disruptive solutions in the detection of HLB: A. The use of the Agrowing’s sensors for confirmation of inspectors’ visual observation (increasing their accuracy), and B. Early aerial detection of HLB infected trees by a UAV (drone).

VetorGEO has already initiated an aerial research to verify the feasibility of detecting HLB infected trees from 60m altitude. Tests carried out in orchards with advanced symptoms of the disease showed promising results. The challenge now, is to detect the initial symptoms of the disease.

VetorGEO is conducting these very days PCR tests of selected trees, suspected to be infected with HLB, based on the multispectral imagery, which was acquired by Agrowing’s 8MP and 12MP per band 10 band sensor. Once the PCR tests will be analyzed it will be possible to assess the accuracy level of such high altitude aerial detection.

Dr. Kelly Pazolini, VetorGEO’s scientific researcher, says that even if the aerial detection will eventually prove to be only as accurate as the professional visual inspection (up to 70% accurate), it will be a huge step forward. “The rapid detection of trees, which are suspected as infected by HLB, in large orchards, will enable growers to conduct PCR tests for such selected trees without delay, enabling fast decision making, reducing production losses, production costs, and even more important - prevention of HLB spreading.”

Today, visual inspections are performed by 4 inspectors on a tractor pulled platform. Well trained inspectors may inspect 2500 to 3000 trees per day. According to VetorGEO, a single UAV equipped with Agrowing’s 10 band 8 or 12 Megapixel per band sensor will scan over 300ha per day, 150,000 - 175.000 trees, not only reducing the cost of the visual inspections dramatically but enabling the desired early detection of such infected trees.

VetorGEO’s research team intends to fine tune its AI algorithms for HLB detection based on the actual PCR results, aiming at as low as possible false negative results and as high as possible percentage of accurate positive detection, even at the expense of having up to 50% false positive results.

Agrowing’s award-winning multispectral sensors are unique in their high-resolution, number of spectral bands (4/10/14), wide dynamic-range and their ability to acquire aligned multispectral data from both high and low altitude.

Agrowing and VetorGeo are interconnected with the Brazilian universities USP and UNESP, in the respective areas of phytopathologies and remote sensing, providing tools and sensors for various scientific projects.

Bayer Crop Science LATAM (BAYERCROP.NS) has recently selected VetorGEO out of 150 companies for a project, where VetorGEO will start in January of 2021 a research exploring the effectiveness of VetorGEO/Agrowing multispectral solution in different crops for Bayer.

“We are naturally happy with the results of VetorGEO’s HLB research.” Says Ira Dvir, Co-Founder and CEO of Agrowing. “Agrowing’s research in potatoes produced incredible results, detecting early Blight and Alternaria with almost no false negative or false positive identification. Growers do not care about technology as such, they need trustworthy actionable data, to control the spreading of pests and diseases like HLB, and our sensors were designed with this goal in mind, taking precision agriculture from the good old NDVI to the next level AI. We encourage researches and service providers to explore new horizons using our 10 and 14 band sensors.”

VetorGEO is a Brazilian company with Israeli participation, focused on developing technologies for digital precision agriculture, providing solutions for agricultural diagnosis. VetorGEO is focus on citrus, collaborating in scientific research groups with the leading Brazilian universities Usp and Unesp.

The Israeli Agrowing, founded in 2014, collaborates with academic institutes around the globe, like the Rochester Institute of Technology, the Israeli Technion and Unesp, with the goal of enhancing its line patented products and image acquisition technology, to provide service providers with the best multispectral sensor and methods for fast and accurate AI identification of pests, diseases, and nutrient deficiencies.

In October 2019, Agrowing's Quad-lens sensor won the press award as one of the three innovative products of the Commercial UAV Expo Americas.