SlideShare a Scribd company logo
1 of 49
Download to read offline
Automatic planogram recognition solution
by .
IdealoRama
Why do you need it?
Did you know what is a planogram?
By analyzing past and current buying patterns, a planogram
manager can make recommendations about the number of
"facings" (product faces) a certain product should have on a retail
display. The diagram will indicate a certain (shelf) height level
from the floor on which the product should be displayed, as well
as which products should surround it.
Moreover product location within a store is important
for supplier, who is interested in the most favorable and
crowded place in order to increase sales.
A planogram is a diagram that shows how and where specific retail products should be
placed on shelves or displays. The main motive to have a planogram is to ensure the convenient and logical
location of products on a retail shelf in order to increase sales.
Today
Conventional planogram management
Thousands of employees
check products on-shelf availability
and out-of-stock manually,
validate correctness of disposition
with the requirements and
agreements. This takes time,
money, and mistakes are very
likely.
Let's thinking what do we have now..
Many manufacturers enter into agreements with retailers to display their
goods in certain ways. Everyone wants their product to be placed in the center opposite
consumers eyes and occupy the largest possible area on a shelf. But this seldom
works as expected.
Thousands of merchandisers move from store to store to make sure that their
company's products are on the shelf and presented in accordance with the contract.
Sometimes they are lazy: it is much easier to forge the report without leaving
your home than to go to a point of sale.
Next
The future of planogram management
Our complete solution now
equipped with AI helps
optimize and improve all
processes of planning, building and
controlling planograms.
It enables you to control the
placement of thousands of products
with little or no human
involvement.
Let's imagine what is needed for a store ..
We select products directly from the database, simply by ticking the checkbox opposite
the required product in the database. Instantly, all selected positions are pulled into the
constructor, in the form of rectangles with their own proportions and sizes. In the
planogram designer software, we set the sizes of the available shelves and arrange the
virtual products on them. The planogram is ready! Immediately, we send it via online to a
responsible employee. He receives an understandable and clear layout to implement on
the shelves. The next day, an employee takes pictures of the actual shelves and in case
there are an error you’ll immediately receive a notification!
We are
Let we introduce ourself
Since 2006 our R&D office in Eastern Europe, Ukraine has developed dozens of projects
for retail which nowadays operate in Ukraine, Russia, Kazakhstan, EU and the US.
We have an extensive background in CSP (content service providers), PIM (product
information management), EDI (electronic data interchange), SCM (supply chain
management) systems.
The idea of Idealorama originally comes from repeated inquiries from our regular retail
customers. The IdealoRama turned out so cool that we want to offer it to you!
We proud to have such successful partners
We made the
`Next` closer!
What do we offer
These are 3 software modules
These three modules are ideally combined with each other and make up the whole
ecosystem. But you can use any component individually or combining it with another
component.
The complex consists of 3 main modules:
Listex — is a digital catalog of
consumer goods. Listex.info is an
open catalogue that can be used by
retailers, suppliers, distributors, and
customers.
Product catalogue module
The complex consists of 3 main modules:
This is online WYSIWYG designer for
creating planograms. Plexa pulls
products data from Listex. Created
planograms can be printed out, stored or
sent as a PDF file by mail and are
immediately available to any employee
responsible for placing products on a
shelf.
Planogram recognition module
The complex consists of 3 main modules:
This module is used for automated
verification of products on a shelf, and
validate their positions against
planogram.
It requires moderate AI training (to
recognize your actual products) before
the neural network will be able to
identify the products on the picture.
Planogram management module
You can use it almost in any configuration
(no need to replace your existing MDM or planogram management software)
IdealoRama
The modular software complex to enable planogram
compliance within your organization
Step 0 Starting screen
Shown for a few seconds while app is loading.
We require app to be vertical layout only, ready for
iphone8 and above.
Step 1 Camera capture
Camera capture is shown the way it eagerly uses all
available screen space. It’s overlayed with the
button to make a shoot which is placed in the
middle of the bottom line.
While camera capturing, algo trying to mark
products it sees with half-transparent yellow boxes.
Once the user is satisfied he’d tap the shoot button.
Step 2 Image processing
Once “shoot” is pressed, camera capture stopped,
the screen now shows taken image with yellow
semi-transparent boxes on it. The whole screen is
overlayed with a semi-transparent gray box with
“Processing…” message in it.
The app sends taken image to the backend for
processing. Backward the app gets JSON with
coords and product description for each zone.
Step 3.1 Results report visualization
As soon as the app gets a response from the backend, it
should change the layout to show the back button at the top,
and two tabs at the bottom line.
The body must render taken image with semi-transparent red
(product not identified) and green (product identified) boxes.
Left tab should be active.
The image along with the overlayed boxes should be
zoomable with a two-fingers “zoom” gesture. The minimum
zoom level equals to when whole image fits the screen. The
maximum - 5x zoom.
Step 3.2 Results report listing
On the second tab, the user can see a response
from backend represented as a list of products.
Each product represented with it’s cut from taken
image plus product headline and barcode to the
right.
The list should be scrollable.
IdealoRama helps optimize and improve all processes of planning, building and
controlling planograms.
It is modular software for planogram compliance management. Allows you to manage
thousands of SKUs with little or no human involvement.
Automatic planogram recognition solution IdealoRama
Planning and Building aspect
The functional has a set of mock-ups of equipment of a store with dimensions of
racks and shelves.
Thanks to the Listex digital catalogue, we have photos and weight & size
characteristics that are used in the construction of the planogram.
The ability to save the finished planogram in PDF format and send directly to the
store.
Controlling aspect
Special robotic device scans a shelf with a display of products.
The shelf photo is sent to the server for analysis by specially trained machine
learning algorithms.
Algorithms recognize products on the photo based on their distinctive features
which combine the following parameters: logo, color, shape, type of packaging.
1
2
3
Controlling aspect
From the photo of the shelf with the goods, IdealoRama automatically determines
which goods are present in the display and whether the display corresponds to
the planogram, and creates an analytical report.
IdealoRama automatically recognizes goods (SKU) on a shelf by the machine
learning methods. Thus, now it is not necessary for the merchandiser to
photograph the display on the shelf and upload photos for further manual
processing.
4
Planogram compliance control with IdealoRama
IdealoRama automatically determines from the photograph or video of the shelf
with the goods which goods are present in the display and whether the display
corresponds to the planogram, and then creates an analytical report.
Analysis is performed automatically solely on the basis of incoming pictures and
does not require additional calculations. The report is transmitted to the
company’s department (or to responsible employee) via mobile application. The
response from the system comes in real-time, which means that the employee
can take actions immediately.
IdealoRama’s
Customer benefits
What do you get
IdealoRama allows you
to manage thousands
of products with
virtually no human
involvement.
Being fully automated,
IdealoRama minimizes
the human factor when
collecting and analyzing
information, reducing
the percentage of errors
close to zero.
Recognition accuracy
equal or greater then
95%
(depends on quality of learning dataset)
Merchandisers can deal
with sales and
promotions of goods,
rather than constantly
monitoring the display
of goods on a shelf.
Real-time analysis of
any shelf in your
stores.
Online reporting makes
it possible to take
needed actions in a
store immediately.
Received reports allow
you to analyze sales,
orders and stocks of
goods.
You will become
innovative :)
Training your AI
As easy as 1-2-3!
Our neural network is pretrained, but anyway it should be
adopted to your local conditions and it must “know” your
products.
1
We take about 3 000 - 10 000 pictures of your shelves with your products in
several different locations. It should be ~20-50 SKU per shot which is
approximately 500 000 assigned barcodes in total.
Then we identify each product on the pictures, assign a class (one of box, pack,
bag, bottle) and set it’s barcode (with help of LISTEX product catalogue or your
own existing MDM system). It is the most time-consuming stage of the project
called data-acquisition.
1
2
After such actions taken, the train dataset is ready to be processed, we call it
training. It will take about 10-20 days and will require a computer with hi-end
video card (NVIDIA Titan RTX, NVIDIA Quadro and similar) having powerful GPU
and huge amount of video-memory (apx. 24GB+).
3
Quite a lot of mistakes and strange features happen in the process of training
and
we might have to run through do-check-fix-restart loop several times
Training procedure -1-
Train dataset photo-session
Taking 10k+ photos of racks in 10 trade halls
Training procedure -2-
Photo annotation
Annotating images - identifying each product by record it’s class (box,
bottle, can, bag, etc.) and product identifier (UPC, barcode, or article
number)
Training procedure -3-
DNN training
Process of DNN-training is happening (~10-20 days)
Training procedure -4-
Calculate descriptors
Just born AI calculates descriptors which will help to recognize any
products in your database
Training procedure -5-
Production detection
New SKU coming
We can outsource training process
for your projects as well
Now we have well-trained Alpha.hdf5!
Our team comes to a store and takes pictures of product shelves using different
smartphones (with different cameras).
Annotation team identifies products on these pictures (one rectangle box = one
product).
Product detection neural network (PDNN) gets trained to identify these products.
When predefined loss function has no longer decreasing and test & validation are
going well, learning process stops.
1
2
3
4
IdealoRama’s
magic sauce
IdealoRama an inside look
2 Products detection process
Report
3 Identification 4 Image descriptor
5 Ranking
descriptors library
1 Taking pictures of a shelf
Magic sauce
1 Take pictures 2 Products detection on a shot
Weights file for detection
(Alpha.hdf5)
Magic sauce
3 Identification
Weight file for auto encoder
(Golf.hdf5)
4 Image descriptor
(4096 x 3 byte floating
point real numbers)
Magic sauce
5 Ranking
Ranking by sum of finite differences
of pixel subtraction of the first or
second order
Report
Descriptors library
JSON with
product barcodes
and boxes
Thank you for your attention!
https://idealorama.com
https://listex.online
+1 (415) 890-20-97

More Related Content

Similar to Listex - IdealoRama - Plexa

Rick Watkins Power Point presentation
Rick Watkins Power Point presentationRick Watkins Power Point presentation
Rick Watkins Power Point presentationrickwatkins
 
Engineering plant facilities 17 artificial intelligence algorithms & proc...
Engineering plant facilities 17 artificial intelligence algorithms & proc...Engineering plant facilities 17 artificial intelligence algorithms & proc...
Engineering plant facilities 17 artificial intelligence algorithms & proc...Luis Cabrera
 
Engineering plant facilities 17 artificial intelligence algorithms & proc...
Engineering plant facilities 17 artificial intelligence algorithms & proc...Engineering plant facilities 17 artificial intelligence algorithms & proc...
Engineering plant facilities 17 artificial intelligence algorithms & proc...Luis Cabrera
 
[[Srs]] online shopping website for TYBSC IT
[[Srs]] online shopping website for TYBSC IT[[Srs]] online shopping website for TYBSC IT
[[Srs]] online shopping website for TYBSC ITYogeshDhamke2
 
Pulsar magento responsive theme magik
Pulsar magento responsive theme magikPulsar magento responsive theme magik
Pulsar magento responsive theme magikNETBASE CMSMART
 
Rick Watkins Power Point Presentation on Automation efficiencies
Rick Watkins Power Point Presentation on Automation efficienciesRick Watkins Power Point Presentation on Automation efficiencies
Rick Watkins Power Point Presentation on Automation efficienciesrickwatkins
 
Power Point Presentation
Power Point PresentationPower Point Presentation
Power Point Presentationrickwatkins
 
Create Your eCommerce Photo Catalogue
Create Your eCommerce Photo CatalogueCreate Your eCommerce Photo Catalogue
Create Your eCommerce Photo CatalogueeTailing India
 
Create Your eCommerce Photo Catalogue
Create Your eCommerce Photo CatalogueCreate Your eCommerce Photo Catalogue
Create Your eCommerce Photo CatalogueAshish Jhalani
 
Artificial Intelligence Services - AAPNA Infotech
Artificial Intelligence Services - AAPNA InfotechArtificial Intelligence Services - AAPNA Infotech
Artificial Intelligence Services - AAPNA InfotechAapna Infotech
 
SAP CMMS Software Productoo 2018 | Smart Factory
SAP CMMS Software Productoo 2018 | Smart FactorySAP CMMS Software Productoo 2018 | Smart Factory
SAP CMMS Software Productoo 2018 | Smart FactoryProductoo Software
 
Production Kanban Software Productoo 2018 | Smart Factory
Production Kanban Software Productoo 2018 | Smart FactoryProduction Kanban Software Productoo 2018 | Smart Factory
Production Kanban Software Productoo 2018 | Smart FactoryProductoo Software
 
ValueClothing, SAP, Extended Warehouse Management Proposal
ValueClothing, SAP, Extended Warehouse Management ProposalValueClothing, SAP, Extended Warehouse Management Proposal
ValueClothing, SAP, Extended Warehouse Management ProposalGeorgios Daskalakis
 
MongoDB World 2018: Building Intelligent Apps with MongoDB & Google Cloud
MongoDB World 2018: Building Intelligent Apps with MongoDB & Google CloudMongoDB World 2018: Building Intelligent Apps with MongoDB & Google Cloud
MongoDB World 2018: Building Intelligent Apps with MongoDB & Google CloudMongoDB
 
MongoDB.local Austin 2018: Building Intelligent Apps with MongoDB & Google Cloud
MongoDB.local Austin 2018: Building Intelligent Apps with MongoDB & Google CloudMongoDB.local Austin 2018: Building Intelligent Apps with MongoDB & Google Cloud
MongoDB.local Austin 2018: Building Intelligent Apps with MongoDB & Google CloudMongoDB
 
INARTS - Mobile Merch Application
INARTS - Mobile Merch ApplicationINARTS - Mobile Merch Application
INARTS - Mobile Merch Applicationferryinarts
 
Building Intelligent Apps with MongoDB and Google Cloud - Jane Fine
Building Intelligent Apps with MongoDB and Google Cloud - Jane FineBuilding Intelligent Apps with MongoDB and Google Cloud - Jane Fine
Building Intelligent Apps with MongoDB and Google Cloud - Jane FineMongoDB
 

Similar to Listex - IdealoRama - Plexa (20)

MG Case Studies 1
MG Case Studies 1MG Case Studies 1
MG Case Studies 1
 
Inventerium
InventeriumInventerium
Inventerium
 
Rick Watkins Power Point presentation
Rick Watkins Power Point presentationRick Watkins Power Point presentation
Rick Watkins Power Point presentation
 
Engineering plant facilities 17 artificial intelligence algorithms & proc...
Engineering plant facilities 17 artificial intelligence algorithms & proc...Engineering plant facilities 17 artificial intelligence algorithms & proc...
Engineering plant facilities 17 artificial intelligence algorithms & proc...
 
Engineering plant facilities 17 artificial intelligence algorithms & proc...
Engineering plant facilities 17 artificial intelligence algorithms & proc...Engineering plant facilities 17 artificial intelligence algorithms & proc...
Engineering plant facilities 17 artificial intelligence algorithms & proc...
 
[[Srs]] online shopping website for TYBSC IT
[[Srs]] online shopping website for TYBSC IT[[Srs]] online shopping website for TYBSC IT
[[Srs]] online shopping website for TYBSC IT
 
Pulsar magento responsive theme magik
Pulsar magento responsive theme magikPulsar magento responsive theme magik
Pulsar magento responsive theme magik
 
Rick Watkins Power Point Presentation on Automation efficiencies
Rick Watkins Power Point Presentation on Automation efficienciesRick Watkins Power Point Presentation on Automation efficiencies
Rick Watkins Power Point Presentation on Automation efficiencies
 
Power Point Presentation
Power Point PresentationPower Point Presentation
Power Point Presentation
 
Create Your eCommerce Photo Catalogue
Create Your eCommerce Photo CatalogueCreate Your eCommerce Photo Catalogue
Create Your eCommerce Photo Catalogue
 
Create Your eCommerce Photo Catalogue
Create Your eCommerce Photo CatalogueCreate Your eCommerce Photo Catalogue
Create Your eCommerce Photo Catalogue
 
Artificial Intelligence Services - AAPNA Infotech
Artificial Intelligence Services - AAPNA InfotechArtificial Intelligence Services - AAPNA Infotech
Artificial Intelligence Services - AAPNA Infotech
 
SAP CMMS Software Productoo 2018 | Smart Factory
SAP CMMS Software Productoo 2018 | Smart FactorySAP CMMS Software Productoo 2018 | Smart Factory
SAP CMMS Software Productoo 2018 | Smart Factory
 
Production Kanban Software Productoo 2018 | Smart Factory
Production Kanban Software Productoo 2018 | Smart FactoryProduction Kanban Software Productoo 2018 | Smart Factory
Production Kanban Software Productoo 2018 | Smart Factory
 
ValueClothing, SAP, Extended Warehouse Management Proposal
ValueClothing, SAP, Extended Warehouse Management ProposalValueClothing, SAP, Extended Warehouse Management Proposal
ValueClothing, SAP, Extended Warehouse Management Proposal
 
MongoDB World 2018: Building Intelligent Apps with MongoDB & Google Cloud
MongoDB World 2018: Building Intelligent Apps with MongoDB & Google CloudMongoDB World 2018: Building Intelligent Apps with MongoDB & Google Cloud
MongoDB World 2018: Building Intelligent Apps with MongoDB & Google Cloud
 
Ecommerce Product Reverse Image Search
Ecommerce Product Reverse Image SearchEcommerce Product Reverse Image Search
Ecommerce Product Reverse Image Search
 
MongoDB.local Austin 2018: Building Intelligent Apps with MongoDB & Google Cloud
MongoDB.local Austin 2018: Building Intelligent Apps with MongoDB & Google CloudMongoDB.local Austin 2018: Building Intelligent Apps with MongoDB & Google Cloud
MongoDB.local Austin 2018: Building Intelligent Apps with MongoDB & Google Cloud
 
INARTS - Mobile Merch Application
INARTS - Mobile Merch ApplicationINARTS - Mobile Merch Application
INARTS - Mobile Merch Application
 
Building Intelligent Apps with MongoDB and Google Cloud - Jane Fine
Building Intelligent Apps with MongoDB and Google Cloud - Jane FineBuilding Intelligent Apps with MongoDB and Google Cloud - Jane Fine
Building Intelligent Apps with MongoDB and Google Cloud - Jane Fine
 

Recently uploaded

The 15 Minute Breakdown: 2024 Beauty Marketing Study
The 15 Minute Breakdown: 2024 Beauty Marketing StudyThe 15 Minute Breakdown: 2024 Beauty Marketing Study
The 15 Minute Breakdown: 2024 Beauty Marketing StudyTinuiti
 
Indian Call Girl In Dubai #$# O5634O3O18 #$# Dubai Call Girl
Indian Call Girl In Dubai #$# O5634O3O18 #$# Dubai Call GirlIndian Call Girl In Dubai #$# O5634O3O18 #$# Dubai Call Girl
Indian Call Girl In Dubai #$# O5634O3O18 #$# Dubai Call GirlAroojKhan71
 
The 15 Minute Breakdown: 2024 Beauty Marketing Study
The 15 Minute Breakdown: 2024 Beauty Marketing StudyThe 15 Minute Breakdown: 2024 Beauty Marketing Study
The 15 Minute Breakdown: 2024 Beauty Marketing StudyKatherineBishop4
 
Best VIP Call Girls Noida Sector 51 Call Me: 8448380779
Best VIP Call Girls Noida Sector 51 Call Me: 8448380779Best VIP Call Girls Noida Sector 51 Call Me: 8448380779
Best VIP Call Girls Noida Sector 51 Call Me: 8448380779Delhi Call girls
 
Supermarket Floral Ad Roundup- Week 17 2024.pdf
Supermarket Floral Ad Roundup- Week 17 2024.pdfSupermarket Floral Ad Roundup- Week 17 2024.pdf
Supermarket Floral Ad Roundup- Week 17 2024.pdfKarliNelson4
 
Film= Dubai Call Girls O525547819 Call Girls Dubai Whsatapp
Film= Dubai Call Girls O525547819 Call Girls Dubai WhsatappFilm= Dubai Call Girls O525547819 Call Girls Dubai Whsatapp
Film= Dubai Call Girls O525547819 Call Girls Dubai Whsatappkojalkojal131
 
Dubai Call Girls O525547&19 (Asii) Call Girls Dubai
Dubai Call Girls O525547&19 (Asii) Call Girls DubaiDubai Call Girls O525547&19 (Asii) Call Girls Dubai
Dubai Call Girls O525547&19 (Asii) Call Girls Dubaikojalkojal131
 
Best VIP Call Girls Noida Sector 50 Call Me: 8448380779
Best VIP Call Girls Noida Sector 50 Call Me: 8448380779Best VIP Call Girls Noida Sector 50 Call Me: 8448380779
Best VIP Call Girls Noida Sector 50 Call Me: 8448380779Delhi Call girls
 
Best VIP Call Girls Noida Sector 55 Call Me: 8448380779
Best VIP Call Girls Noida Sector 55 Call Me: 8448380779Best VIP Call Girls Noida Sector 55 Call Me: 8448380779
Best VIP Call Girls Noida Sector 55 Call Me: 8448380779Delhi Call girls
 
Top Rated Pune Call Girls Talegaon Dabhade ⟟ 6297143586 ⟟ Call Me For Genuin...
Top Rated  Pune Call Girls Talegaon Dabhade ⟟ 6297143586 ⟟ Call Me For Genuin...Top Rated  Pune Call Girls Talegaon Dabhade ⟟ 6297143586 ⟟ Call Me For Genuin...
Top Rated Pune Call Girls Talegaon Dabhade ⟟ 6297143586 ⟟ Call Me For Genuin...Call Girls in Nagpur High Profile
 

Recently uploaded (10)

The 15 Minute Breakdown: 2024 Beauty Marketing Study
The 15 Minute Breakdown: 2024 Beauty Marketing StudyThe 15 Minute Breakdown: 2024 Beauty Marketing Study
The 15 Minute Breakdown: 2024 Beauty Marketing Study
 
Indian Call Girl In Dubai #$# O5634O3O18 #$# Dubai Call Girl
Indian Call Girl In Dubai #$# O5634O3O18 #$# Dubai Call GirlIndian Call Girl In Dubai #$# O5634O3O18 #$# Dubai Call Girl
Indian Call Girl In Dubai #$# O5634O3O18 #$# Dubai Call Girl
 
The 15 Minute Breakdown: 2024 Beauty Marketing Study
The 15 Minute Breakdown: 2024 Beauty Marketing StudyThe 15 Minute Breakdown: 2024 Beauty Marketing Study
The 15 Minute Breakdown: 2024 Beauty Marketing Study
 
Best VIP Call Girls Noida Sector 51 Call Me: 8448380779
Best VIP Call Girls Noida Sector 51 Call Me: 8448380779Best VIP Call Girls Noida Sector 51 Call Me: 8448380779
Best VIP Call Girls Noida Sector 51 Call Me: 8448380779
 
Supermarket Floral Ad Roundup- Week 17 2024.pdf
Supermarket Floral Ad Roundup- Week 17 2024.pdfSupermarket Floral Ad Roundup- Week 17 2024.pdf
Supermarket Floral Ad Roundup- Week 17 2024.pdf
 
Film= Dubai Call Girls O525547819 Call Girls Dubai Whsatapp
Film= Dubai Call Girls O525547819 Call Girls Dubai WhsatappFilm= Dubai Call Girls O525547819 Call Girls Dubai Whsatapp
Film= Dubai Call Girls O525547819 Call Girls Dubai Whsatapp
 
Dubai Call Girls O525547&19 (Asii) Call Girls Dubai
Dubai Call Girls O525547&19 (Asii) Call Girls DubaiDubai Call Girls O525547&19 (Asii) Call Girls Dubai
Dubai Call Girls O525547&19 (Asii) Call Girls Dubai
 
Best VIP Call Girls Noida Sector 50 Call Me: 8448380779
Best VIP Call Girls Noida Sector 50 Call Me: 8448380779Best VIP Call Girls Noida Sector 50 Call Me: 8448380779
Best VIP Call Girls Noida Sector 50 Call Me: 8448380779
 
Best VIP Call Girls Noida Sector 55 Call Me: 8448380779
Best VIP Call Girls Noida Sector 55 Call Me: 8448380779Best VIP Call Girls Noida Sector 55 Call Me: 8448380779
Best VIP Call Girls Noida Sector 55 Call Me: 8448380779
 
Top Rated Pune Call Girls Talegaon Dabhade ⟟ 6297143586 ⟟ Call Me For Genuin...
Top Rated  Pune Call Girls Talegaon Dabhade ⟟ 6297143586 ⟟ Call Me For Genuin...Top Rated  Pune Call Girls Talegaon Dabhade ⟟ 6297143586 ⟟ Call Me For Genuin...
Top Rated Pune Call Girls Talegaon Dabhade ⟟ 6297143586 ⟟ Call Me For Genuin...
 

Listex - IdealoRama - Plexa

  • 1. Automatic planogram recognition solution by . IdealoRama
  • 2. Why do you need it? Did you know what is a planogram?
  • 3. By analyzing past and current buying patterns, a planogram manager can make recommendations about the number of "facings" (product faces) a certain product should have on a retail display. The diagram will indicate a certain (shelf) height level from the floor on which the product should be displayed, as well as which products should surround it. Moreover product location within a store is important for supplier, who is interested in the most favorable and crowded place in order to increase sales. A planogram is a diagram that shows how and where specific retail products should be placed on shelves or displays. The main motive to have a planogram is to ensure the convenient and logical location of products on a retail shelf in order to increase sales.
  • 5. Thousands of employees check products on-shelf availability and out-of-stock manually, validate correctness of disposition with the requirements and agreements. This takes time, money, and mistakes are very likely.
  • 6. Let's thinking what do we have now.. Many manufacturers enter into agreements with retailers to display their goods in certain ways. Everyone wants their product to be placed in the center opposite consumers eyes and occupy the largest possible area on a shelf. But this seldom works as expected. Thousands of merchandisers move from store to store to make sure that their company's products are on the shelf and presented in accordance with the contract. Sometimes they are lazy: it is much easier to forge the report without leaving your home than to go to a point of sale.
  • 7. Next The future of planogram management
  • 8. Our complete solution now equipped with AI helps optimize and improve all processes of planning, building and controlling planograms. It enables you to control the placement of thousands of products with little or no human involvement.
  • 9. Let's imagine what is needed for a store .. We select products directly from the database, simply by ticking the checkbox opposite the required product in the database. Instantly, all selected positions are pulled into the constructor, in the form of rectangles with their own proportions and sizes. In the planogram designer software, we set the sizes of the available shelves and arrange the virtual products on them. The planogram is ready! Immediately, we send it via online to a responsible employee. He receives an understandable and clear layout to implement on the shelves. The next day, an employee takes pictures of the actual shelves and in case there are an error you’ll immediately receive a notification!
  • 10. We are Let we introduce ourself
  • 11. Since 2006 our R&D office in Eastern Europe, Ukraine has developed dozens of projects for retail which nowadays operate in Ukraine, Russia, Kazakhstan, EU and the US. We have an extensive background in CSP (content service providers), PIM (product information management), EDI (electronic data interchange), SCM (supply chain management) systems. The idea of Idealorama originally comes from repeated inquiries from our regular retail customers. The IdealoRama turned out so cool that we want to offer it to you!
  • 12. We proud to have such successful partners
  • 13. We made the `Next` closer! What do we offer
  • 14. These are 3 software modules These three modules are ideally combined with each other and make up the whole ecosystem. But you can use any component individually or combining it with another component.
  • 15. The complex consists of 3 main modules: Listex — is a digital catalog of consumer goods. Listex.info is an open catalogue that can be used by retailers, suppliers, distributors, and customers. Product catalogue module
  • 16. The complex consists of 3 main modules: This is online WYSIWYG designer for creating planograms. Plexa pulls products data from Listex. Created planograms can be printed out, stored or sent as a PDF file by mail and are immediately available to any employee responsible for placing products on a shelf. Planogram recognition module
  • 17. The complex consists of 3 main modules: This module is used for automated verification of products on a shelf, and validate their positions against planogram. It requires moderate AI training (to recognize your actual products) before the neural network will be able to identify the products on the picture. Planogram management module
  • 18. You can use it almost in any configuration (no need to replace your existing MDM or planogram management software)
  • 19. IdealoRama The modular software complex to enable planogram compliance within your organization
  • 20. Step 0 Starting screen Shown for a few seconds while app is loading. We require app to be vertical layout only, ready for iphone8 and above.
  • 21. Step 1 Camera capture Camera capture is shown the way it eagerly uses all available screen space. It’s overlayed with the button to make a shoot which is placed in the middle of the bottom line. While camera capturing, algo trying to mark products it sees with half-transparent yellow boxes. Once the user is satisfied he’d tap the shoot button.
  • 22. Step 2 Image processing Once “shoot” is pressed, camera capture stopped, the screen now shows taken image with yellow semi-transparent boxes on it. The whole screen is overlayed with a semi-transparent gray box with “Processing…” message in it. The app sends taken image to the backend for processing. Backward the app gets JSON with coords and product description for each zone.
  • 23. Step 3.1 Results report visualization As soon as the app gets a response from the backend, it should change the layout to show the back button at the top, and two tabs at the bottom line. The body must render taken image with semi-transparent red (product not identified) and green (product identified) boxes. Left tab should be active. The image along with the overlayed boxes should be zoomable with a two-fingers “zoom” gesture. The minimum zoom level equals to when whole image fits the screen. The maximum - 5x zoom.
  • 24. Step 3.2 Results report listing On the second tab, the user can see a response from backend represented as a list of products. Each product represented with it’s cut from taken image plus product headline and barcode to the right. The list should be scrollable.
  • 25. IdealoRama helps optimize and improve all processes of planning, building and controlling planograms. It is modular software for planogram compliance management. Allows you to manage thousands of SKUs with little or no human involvement. Automatic planogram recognition solution IdealoRama
  • 26.
  • 27. Planning and Building aspect The functional has a set of mock-ups of equipment of a store with dimensions of racks and shelves. Thanks to the Listex digital catalogue, we have photos and weight & size characteristics that are used in the construction of the planogram. The ability to save the finished planogram in PDF format and send directly to the store.
  • 28. Controlling aspect Special robotic device scans a shelf with a display of products. The shelf photo is sent to the server for analysis by specially trained machine learning algorithms. Algorithms recognize products on the photo based on their distinctive features which combine the following parameters: logo, color, shape, type of packaging. 1 2 3
  • 29. Controlling aspect From the photo of the shelf with the goods, IdealoRama automatically determines which goods are present in the display and whether the display corresponds to the planogram, and creates an analytical report. IdealoRama automatically recognizes goods (SKU) on a shelf by the machine learning methods. Thus, now it is not necessary for the merchandiser to photograph the display on the shelf and upload photos for further manual processing. 4
  • 30. Planogram compliance control with IdealoRama IdealoRama automatically determines from the photograph or video of the shelf with the goods which goods are present in the display and whether the display corresponds to the planogram, and then creates an analytical report. Analysis is performed automatically solely on the basis of incoming pictures and does not require additional calculations. The report is transmitted to the company’s department (or to responsible employee) via mobile application. The response from the system comes in real-time, which means that the employee can take actions immediately.
  • 32. IdealoRama allows you to manage thousands of products with virtually no human involvement. Being fully automated, IdealoRama minimizes the human factor when collecting and analyzing information, reducing the percentage of errors close to zero. Recognition accuracy equal or greater then 95% (depends on quality of learning dataset) Merchandisers can deal with sales and promotions of goods, rather than constantly monitoring the display of goods on a shelf.
  • 33. Real-time analysis of any shelf in your stores. Online reporting makes it possible to take needed actions in a store immediately. Received reports allow you to analyze sales, orders and stocks of goods. You will become innovative :)
  • 34. Training your AI As easy as 1-2-3!
  • 35. Our neural network is pretrained, but anyway it should be adopted to your local conditions and it must “know” your products. 1 We take about 3 000 - 10 000 pictures of your shelves with your products in several different locations. It should be ~20-50 SKU per shot which is approximately 500 000 assigned barcodes in total. Then we identify each product on the pictures, assign a class (one of box, pack, bag, bottle) and set it’s barcode (with help of LISTEX product catalogue or your own existing MDM system). It is the most time-consuming stage of the project called data-acquisition. 1 2
  • 36. After such actions taken, the train dataset is ready to be processed, we call it training. It will take about 10-20 days and will require a computer with hi-end video card (NVIDIA Titan RTX, NVIDIA Quadro and similar) having powerful GPU and huge amount of video-memory (apx. 24GB+). 3 Quite a lot of mistakes and strange features happen in the process of training and we might have to run through do-check-fix-restart loop several times
  • 37. Training procedure -1- Train dataset photo-session Taking 10k+ photos of racks in 10 trade halls
  • 38. Training procedure -2- Photo annotation Annotating images - identifying each product by record it’s class (box, bottle, can, bag, etc.) and product identifier (UPC, barcode, or article number)
  • 39. Training procedure -3- DNN training Process of DNN-training is happening (~10-20 days)
  • 40. Training procedure -4- Calculate descriptors Just born AI calculates descriptors which will help to recognize any products in your database
  • 41. Training procedure -5- Production detection New SKU coming
  • 42. We can outsource training process for your projects as well
  • 43. Now we have well-trained Alpha.hdf5! Our team comes to a store and takes pictures of product shelves using different smartphones (with different cameras). Annotation team identifies products on these pictures (one rectangle box = one product). Product detection neural network (PDNN) gets trained to identify these products. When predefined loss function has no longer decreasing and test & validation are going well, learning process stops. 1 2 3 4
  • 45. 2 Products detection process Report 3 Identification 4 Image descriptor 5 Ranking descriptors library 1 Taking pictures of a shelf
  • 46. Magic sauce 1 Take pictures 2 Products detection on a shot Weights file for detection (Alpha.hdf5)
  • 47. Magic sauce 3 Identification Weight file for auto encoder (Golf.hdf5) 4 Image descriptor (4096 x 3 byte floating point real numbers)
  • 48. Magic sauce 5 Ranking Ranking by sum of finite differences of pixel subtraction of the first or second order Report Descriptors library JSON with product barcodes and boxes
  • 49. Thank you for your attention! https://idealorama.com https://listex.online +1 (415) 890-20-97