PROVIDING SMART RETAIL From Self-Checkout Systems to Empowering - PowerPoint PPT Presentation

PROVIDING SMART RETAIL From Self-Checkout Systems to Empowering Sales Force IOT Vision AI

Agenda 2  Introduction  Business Use Case  Solution Introduction  Process lifecycle of a on-boarding a customer  Process lifecycle of solution  Data Collection strategies  Pipeline of Data Collection  Helix Platform  Tools for Annotation  Algorithms  Detection  Classification – Product, Package & Size  Rejection  Size Estimation  Transfer Learning  GPU’s  Training  Inferencing  Edge Solutions  Algorithms  GPU’s for Inferencing  Scalability

3 Sell anywhere, Anytime!

4 Using checkout free, self serve assets

Using checkout free, 5 self serve assets

6 Improve your market execution

Leveraging instant actionable insights 7 Retail Shelf/ 1 2 3 4 Outlet Located (GPS) Shelf/Cooler Analyzed Prescriptive Analytics Cooler Captured Simple Image Capture Instant Result Actionable Insights Select outlet with Quality Check processing & Core KBI’s

Shelf and Position level 8 Information UFO-Tetra-0ml UFO-Tetra-0ml Mm_Apl-Tetra-150ml Mm_Apl-Tetra-150ml Maaza_Refresh-Tetra-150ml Maaza_Refresh-Tetra-150ml Rimzim_Spicy-Pet-250ml Rimzim_Spicy-Pet-250ml Sprite-Rgb-200ml Sprite-Rgb-200ml Sprite-Rgb-300ml Sprite-Rgb-300ml Sprite-Rgb-300ml Sprite-Rgb-200ml UFO-Pet-0ml UFO-Pet-0ml Sprite-Pet-600ml Sprite-Pet-475ml Sprite-Pet-400ml Sprite-Pet-400ml UFO-Tetra-0ml UFO-Tetra-0ml UFO-Tetra-0ml Maaza-Pet-1.5L Maaza-Pet-1.5L Maaza-Pet-1.5L Kinley-Pet-1L Kinley-Pet-1L Thums_Up-Pet-750ml Fanta-Pet-2.25L Thums_Up-Pet-1.25L Thums_Up-Pet-1.25L Sprite-Pet-2.25L

9 Remotely know how you products are performing Using our Stick-n-Play IOT and Machine Vision Solutions

Enabling Scale via our Instant Product Cataloging 10 Technology < 3 Minutes to Capture Products @ Source Distribution Factories Centers ` < 24 Hours Start Recognizing New Products

Retail Customer Onboarding – Process Lifecycle 10 PROJECT MANAGEMENT - CUSTOMER ONBOARDING Backend IR validation Phase: Execution and Monitoring IR Algorithms : Model Generation Phase IR Algorithms : Refinement and Enhancement Phase 30 40 45 75 100 55 Infrastructure : Deployment Phase No Completion of On-boarding First Model released Updation of Model Product Onboarding Validation and Model completion for field testing with Field Data Generation is Intiated - Master data sheet - Availability of Field Data - Availability of Field Data - Availability of Field Data - Video and Image upload - Planning of Infrastructure - Completion of to helix data Infrastructure deployment - HD Videos - Field Images if available - Project Management - Project Management - Project Management - Project Management - Project Management Team Team Team Team Team - Programmer- 1 no. / - Programmer- 1 no. / - Programmer- 1 no. / - Programmer- 1 no. / - Programmer- 2 no. / Bottler Bottler Bottler Bottler Bottler - 3 Associates for Execution - 3 Associates for Execution - 3 Associates for Execution - 3 Associates for Execution at Operations at Operations at Operations at Operations - 1 TL for Monitoring at - 1 TL for Monitoring at - 1 TL for Monitoring at - 1 TL for Monitoring at Operations Operations Operations Operations - Infrastructure expert - Infrastructure expert - Application development - Application development expert if necessary expert if necessary

Maxerience AI Solution – Process Lifecycle 11 PROCESS Nos. PROCESS START Master Data Sheet to be PROCESS 1 shared Pictures/ Images of the PROCESS 2 Product in the Helix tool to be carried out and the information to be shared Downloading Pictures / PROCESS 3 Images from Helix portal Verificationof Manual Grouping of Not OK Master sheet PROCESS 4 images based on SKU's & downloaded images OK Obtain HD videos from customer Program run to obtain 800 to 1000 snaps among the Obtain fi PROCESS 5 available snaps for creation image of a Model HD videos to be processed in detector algorithm Process PROCESS 6 Generation of Models for Generation of Models for Classifiers Rejectors detector (Plano Iterative Process for Model Generation Cleaning of Images Cleaning Validationof < 99% Classifier Model PROCESS 7

Data Collection Strategies – Pipeline 12 PROCESS Nos. PROCESS START Master Data Sheet to be PROCESS 1 shared Pictures/ Images of the PROCESS 2 Product in the Helix tool to be carried out and the information to be shared DCS-1 Downloading Pictures / PROCESS 3 Images from Helix portal Verificationof Manual Grouping of Not OK Master sheet PROCESS 4 images based on SKU's & downloaded DCS-2 DCS-3 images OK Obtain HD videos from customer Program run to obtain 800 to 1000 snaps among the Obtain field cooler PROCESS 5 available snaps for creation images from of a Model HD videos to be processed in detector algorithm Process through PROCESS 6 Generation of Models for Generation of Models for detector algorithm Classifiers Rejectors (Planogram) Iterative Process for Model Generation Cleaning of Images Cleaning of Images

Data Collection Strategies – Helix – Smart Phone 13

Data Collection Strategies – Helix – Auto Turntable 14

Data Collection Strategies – Tools for Annotation 15

Scalability 20  Self Service Portal for  On-boarding products  Training products  Validation of Inferences/Results  Gamification  Plug-n-Play Smart Shelves

Monetize retail IoT and Image Recognition data 18 Out- of-the box Self-service actionable insights Monetization of retail IoT and Image Recognition data Machine Key KBI’s Learning

Leveraging IOT & AI for retail 19 At Scale China Bulgaria Italy India Mexico Argentina 5 Countries

20 VISIT US FOR LIVE DEMOS BOOTH – 7102 INNOVATION LAB- 4 TH FLOOR Pradeep V Pydah | CEO 1-510-8960953 pradeep.v.pydah@maxerience.com 22994, Lavender Valley Ct, Ashburn, Virginia 20148