School of Mechanical Engineering Seminar
Monday April 17.4.2023 at 14:00

Wolfson Building of Mechanical Engineering, Room 206

Xylella Diseases detection in orchards grown under nets using UAV-based images

Yasmin Mazor

 Msc student

Tel Aviv University, Institute of Agricultural Engineering, ARO

Supervisors: Dr. Victor Alchanati, Dr.Avishai Sintov

Almond trees are infected by the bacterium Xylella fastidious and the pathogen is vectored by xylem-

feeding sharpshooters and spittlebugs. Currently, no effective management techniques prevent trees

from becoming infected. The Xylella bacterium is introduced into the tip tubes of the tree by a tip-

sucking insect. As a result, there is a disruption in the fluid passage, and eventually, the tree dies. The

almond leaf scorch disease (ALS) was first identified in Israel In 2016 in the almond groves in the north

of the country. The bacterium poses a significant threat to almonds and other crops in Israel, such as

vines, olives, deciduous trees, and citrus fruits. Symptomatic trees that did not die in the previous

season will grow asymptomatic leaves in the following season, but with entering the summer months

the symptoms of the disease will reappear.

Remote sensing is a type of geospatial technology that samples emitted and reflected

electromagnetic (EM) radiation from the Earth’s terrestrial, atmospheric, and aquatic ecosystems

aiming to detect and monitor the area component without any physical contact. The data is collected

by aircraft-based and satellite-based sensor technologies, either passive or active Passive sensors,

which are the sensors that are used in this research, respond to external stimuli, gathering radiation

that is reflected or emitted by an object or the surrounding space. Popular examples of passive

remote sensors include charge-coupled devices, film photography, radiometers, and infrared. Active

sensors use internal stimuli to collect data, emitting energy to scan objects and areas at which point a

sensor measures the energy reflected from the target.

The importance of using remote sensing can be manifested when collecting data from dangerous or

inaccessible areas, with growing relevance in modern society.

Another advantage is that collecting data with remote sensing tools provides fast and repetitive

coverage of extremely large areas for everyday applications.

Front-line remote sensing tools, coupled with artificial intelligence (AI) technologies, have a significant

role in crop monitoring and disease surveillance. Crop-type classification is an example of using

remote sensing to provide precise, timely, and cost-effective information at different spatial,

temporal, and spectral resolutions.

Today, expert agronomists detect the XF disease by scouting. Each orchard is large, e.g, Sde

Nehemiah Almond orchard is about 16 ha.

As a result of the large area it is very hard for the scout to reach optimal detection of the infected

trees, he can even miss infected trees and as a result, the Xylella pathogen will spread widely in the

orchards. The long-term goal of this research is to develop a system that detects and locates infected

trees in an orchard and directs the scout to that trees without any need to search each tree whether

it is infected or not.

The objective of this work was to develop machine learning and classification algorithms that detect

and locate almond trees infected by Xyllela, using images acquired by a UAV.

Join Zoom Meeting

https://tau-ac-il.zoom.us/j/86497933118

פרטים יפורסמו בהמשך

  You are invited to attend a lecture 

Join Zoom Meeting

https://tau-ac-il.zoom.us/j/89506193877?pwd=TnYvTUc5b3I0eXVYNkJINWUzSjZHdz09

COMMUTE-GAN: COMPETETITIVE, MULTIPLE, EFFICIENT GENERATIVE ADVERSARIAL NETWORKS

 By:

 Omkar Joglekar

 MSc student under the supervision of Prof. Goren Gordon

Abstract

 In complex creative scenarios, co-creativity by multiple agents offers great advantages. Each agent has a specific skill set and a set of abilities, which is generally not enough to perform a large and complex task single-handed.
These kinds of tasks benefit substantially from collaboration. In deep learn-ing applications, data generation is an example of such a complex, poten-tially multi-modal task. Previous Generative Adversarial Networks (GANs)focused on using a single generator to generate multi-modal datasets, which is sometimes known to face issues such as mode-collapse and failure to converge. Single generators also must be very large so that they can generalize com-plex datasets, so this method can easily run into memory constraints. The current multi-generator-based works such as MGAN, MMGAN, MADGAN and AdaGAN either require training a classifier online, the use of complex mixture models or sequentially adding generators, which is computationally complex. In this work, we present a simple, novel approach of training com- petitive multiple efficient GANs (ComMutE-GANs), with multiple genera- tors and a single critic/discriminator, without introducing external complex- ities such as a classifier model. We introduce a new component to the genera- tor loss during GAN training, based on the Total Variation Distance (TVD). Our method offers a robust, stable, memory efficient and easily parallelizable architecture. We present a proof-of-concept on the MNIST dataset, which has 10 modes of data. The individual generators learn to generate different digits from the distribution, and together learn to generate the whole dis- tribution. We compare ComMutE-GANs with larger single-generator GANs and show its memory efficiency and increased accuracy.