3. 3
Introducing Winnow
MARC ZORNES
Founder
KEVIN DUFFY
Founder
Top-rated Manager at McKinsey
with 10 years of food and
sustainability experience
INSEAD MBA with investment
banking
and management consultancy
background
13. Improved ease
of use and
reduced need
for training
Accurate data,
validated by
image
recognition
Automated data
collection
reducing human
error
What if you could use Artificial Intelligence to
capture all food waste data?
13
15. Background to AI
Artificial Narrow
Intelligence (ANI)
Machine Learning Probabilistic methods, ...
Computer Vision
Unsupervised Learning
Supervised Learning
(on big data)
Reinforcement Learning
Linear regression, Supervised Learning (on
moderate data)...
Deep Learning / Neural
Networks
Natural Language
Processing, ...
Artificial Intelligence
Artificial General
Intelligence (AGI)
?
16. Computer Vision is a fast growing form of
Artificial Intelligence
2011 Humans 2018
26% Errors 5% Errors 3% Errors
*Based on Imagenet Challenge
17. 1717
Examples of Computer Vision use cases
• Visual inspection (are there scratches on this phone screen - discard from production line)
• Face recognition (identifying you to open up your phone)
• Analyzing medical imaging to identify health concerns (assist radiologists)
• Self driving cars (object detection, path prediction…)
We are one of very few known companies today using
Computer Vision in kitchens to identify and track food waste.
18. Brining Computer Vision into the
Commercial Kitchen
18
20152013 Today2017
Game changing AI
performance
Computer Vision
compute at the edge
Winnow at scale to
collect data
Birth of in-kitchen
automation
19. 19
Introducing
● Winnow Vision brings the power of AI into
to the kitchen for the first time
● In-built motion sensor camera
automatically captures food thrown away
● Intuitive screen creates engagement with
users via Winnow’s android app
● Digital scale automatically captures weight
of food wasted
● Powered by Nvidia Jetson TX2 – the
world’s fastest embedded AI computing
device
● Connects to Winnow’s cloud analytics
platform to generate powerful insights for
teams
20.
21. 21
Winnow Vision
What’s included?
B. Vision Box
Processes images in real time.
Connects unit to cloud analytics platform
D. Scale
Accurately captures weight
C. Internet connected tablet
Menu configured to kitchen’s
on Winnow application
A. Motion sensitive camera
Automatically captures images
A
BC
D
22. 22
Live output from Winnow Vision model delivering
“predictions”
BEFORE AFTER
DETECTED CHANGE OVERLAID CHANGE
BACKGROUND
FISH AND CHIPS
BACKGROUND
PIRI PIRI CHICKEN
BACKGROUND
ALMOND CAKE WITH DAIM
24. 24
How does Winnow help reduce food waste?
Record all food
waste through the
Winnow system
Use daily and
weekly reports to
identify top areas of
waste
Discuss and
implement changes
as a team
Winnow sites
reduce waste by
40-70%
Record Waste Track Waste Make Changes Reduce Waste
25.
26. 26
Food waste reductions typically happen in 3 phases
1
Low hanging fruit &
quick wins
2
Identify further
efficiency gains
3
Maintaining long-term
reductions
Avoided cost of food waste
Average reduction 55%
Food waste reduction % from baseline over 1 year
(data recorded from 700 Winnow sites)
27. Winnow Financial Benefit – Thailand Hotels by Segment
Source: Winnow reports
Segment
% of Net Food Sales* Waste Value (THB)
Baseline March avg. % Change Baseline March avg. % Change
Midscale 3.38% .94% -72.16% 43,142 26,054 -39.61%
Upscale 2.52% 1.51% -39.98% 35,490 21,296 -39.99%
Luxury 2.59% .81% -68.65% 45,921 20,419 -55.53%
3.38%
2.52% 2.59%
0.94%
1.51%
0.81%
0.00%
2.00%
4.00%
Midscale Upscale Luxury
% of Net Food Sales
Baseline March avg.
43,142
35,490
45,921
26,054 21,296 20,419
-
20,000
40,000
60,000
Midscale Upscale Luxury
Waste Value (THB)
Baseline March avg.
*% of Net Food Sales= Total Food Waste Value/Total Food Sales
Data Based on 25 Live Winnow sites in Thailand
28. Segment
Weight (KG)* Grams/Cover
Baseline March avg. % Change Baseline March avg. % Change
Midscale 807 1334 65.22% 132 91 -30.63%
Upscale 580 340 -41.44% 190 114 -39.81%
Luxury 595 712 19.66% 168 152 -9.58%
Winnow Environmental Benefit – Thailand Hotels by Segment
Source: Winnow reports
*Weight= All forms of Food Waste- Spoilage, Preparation, Overproduction, & Plate Waste
Weight increases due to fluctuations in Plate Waste Figures
Data Based on 25 Live Winnow sites in Thailand
807
580 595
1334
340
712
0
500
1000
1500
Midscale Upscale Luxury
Weight (KG)
Baseline March avg.
132
190 168
91
114
152
-
100
200
Midscale Upscale Luxury
Grams/Cover
Baseline March avg.