~A Deep Learning Model for Post Harvest Grape
Quality Forecasting
Yotam Givati, M.Sc. student at the Department of Industrial Engineering at Tel Aviv University
The lecture will beheld on
Tuesday, October 19, 2021, at 14:00
Via Zoom
https://tau-ac-il.zoom.us/j/81388449216?pwd=QU91L0pXVHc0dS90bFZaUjBoS1Fk...
recording
~https://tau.cloud.panopto.eu/Panopto/Pages/Viewer.aspx?id=ad2bfc85-2d1d-...
Abstract:
Preventing food waste has always been major objective in food distribution and the food retail industry. Yet, it is
estimated that in the US about 40\% of the food supply or 36 million tons are being wasted. When it comes to
storage of fresh produce like fruits and vegetable, the leading distribution method is First In First Out (FIFO). In
FIFO the produce is marketed according to its storage time regardless of any other parameter at storage time.
However, recent advances in digitization, data collection and machine learning allow for improvements in
traditional strategic and planning methodologies.
In this research, we present a novel model for predicting table grapes quality prior to storage. The proposed model
predicts grapes quality based on different features that are measured right after the harvest such as rachis score,
berries firmness, sugar content level, weight, storage temperature, color, etc. Our model utilizes a neural network in
order to map these features into a latent dense representation that removes redundancies while preserving
information related to the future quality of the produce. Then, a multi-objective prediction task is perform in order
to estimate several criteria of grapes quality. Our evaluations showcases the superiority of the proposed model w.r.t
state-of-the-art alternatives.
Bio:
Yotam Givati is an MSc student at the department of Industrial Engineering at Tel-Aviv University, who is writing
a thesis under the supervision of Dr. Noam Keonigstein. Yotam holds a B.sc degree in Industrial Engineering and
management sciences from Tel Aviv University. Previously, Yotam worked as a BI Analyst at Google (Waze),
helped to manage data storage and preform campaigns and feature analysis, participated in designing and execute
research experiments
