SlideShare a Scribd company logo
How to fully
automate a store
A Made in Italy use case
The Team
Alessio Elmi
Artificial Intelligence Engineer
Shruti Verma
Artificial Intelligence Engineer
Michele Toni
Artificial Intelligence Engineer
Bruno Abbate
Machine Learning Engineer
linkedin.com/in/alessioelm
i
linkedin.com/in/shrutiverma2
linkedin.com/in/bruno-abbate
linkedin.com/in/michele-toni
Naser Derakhshan
Computer Scientist
linkedin.com/in/naser-derakhshan-51951828
Pietro Tortella
Mathematician
linkedin.com/in/pietro-tortella-976839ab
Luca Lulleri
Industrial Designer
linkedin.com/in/lucalulleri
Alessandro Re
Machine Learning Engineer
linkedin.com/in/akiross
Riccardo Di Guida
Machine Learning Engineer
linkedin.com/in/riccardo-di-guida-
005764124
Davide Mazzini
Deep Learning Engineer
linkedin.com/in/davidemazzini
Mattia Santachiara
AR/VR Engineer
linkedin.com/in/mattia-santachiara-90a1b379
Igor Moiseev
Crazy CTO
linkedin.com/in/moiseevigor
4 PhD
8 MSc
Automated checkout
What a beast?
Three main ML problems
Object tracking and
Anomaly detection
Pose-estimation and
People Tracking
Assignment problem
It was required a
system which could
validate the correct
amount of goods
picked up or dropped
at the same time from
a user.
Hardware Design
Scales PCB/Firmware
The PICK action
(bottle of water)
The DROP action
(bottle of water)
Camera Positioning Study
1. Retrieve Cad Drawing of the space
2. 3D modeling of the space
3. Define camera position and direction
4. Grasshopper algorithm to make
Camera Positioning Study
Evolutionary Algorithm
and Particle Swarm
Optimization to
optimize camera
positioning.
Pose-estimation
Pose-estimation
Pose-estimation
● Train 2D pose estimation model using a top view dataset
including renderings from the synthetic datasets
● GPU version of the upsampling model (main bottleneck right now)
● Cameras “software” synchronization
● Reduce CPU and GPU load
Tracking
Tracking: The problem
Match 2D-pose
detections from
different cameras to
create 3D-tracks.
Tracking: The glossary
Detection
One pose
in a given frame
at a given time
Reconstruction
Many detections
different frames
at a given time
Track
Many reconstructions
at different times
Tracking: The approach
Hypergraphs for Joint Multi-view Reconstruction and Multi-object Tracking
by M. Hofmann, D. Wold, G. Rigoll. 2013
Tracking: The approach
● Construct all possible
reconstructions and links
● Associate probabilities to them
● Associate probabilities to links
● Create Hypergraph
● Reduce to BIP problem ● Boolean variable per vertex
● Boolean variable per link
● Two constraints per vertex
○ Incoming flow = vertex variable value
○ Outgoing flow = vertex variable value
● Additional constraints from detections
○ Each detection might belong to at most
one flow
● Cost per vertex variable from reconstruction
prob
● Cost per link variable from link prob
● Minimize cost of flow
Binary Integer Programming
Minimize cost with integer variables
satisfying given constraints
Tracking: The approach online
Window 0 Window 2
Window 1 Window 3
● Stabletracks
● Flexible
● No ID switch
● CPU Expensive
● Complexity
● Sensible to parameters calibration
● BIP is NP-Hard
Tracking: The doing
Introduced the 3D geometry of the store.
● Use geometric informations on cameras and obstacles to filter reconstructions
● Make all parameters position-dependent
RESULTS:
➔ Lighter graph (-50% variables, -20% equations)
➔ Reduced complexity → Better scalability to bigger stores
Object tracking and Anomaly detection
� Detect misplaced products in the
scales
� Detect extraneous objects in the
scales
Object tracking and Anomaly detection
1) ODIN (Out-of-distribution detection)
2) Reconstruction-based using Autoencoder
3) Object Detection using ResNet 50 + Faster RCNN
Object tracking and Anomaly detection
DB
Query
image
Input
Reference
image
Resnet-18 Backbone
Resnet-18 Backbone
Shared
weights
Concatenate
Features
Input
Features
Reference
Features
Input
Features
Reference
Object tracking and Anomaly detection
Multilayer Perceptron
Conform
-
anomaly
Output
Assignment problem
Assignment problem
The aim is to combine data from cameras and scales to predict events
e = (timestamp, action, scale, product, quantity, user)
2 INPUT
SOURCES
CAMERAS
SCALES
DATA
PROCESSING
DATA
PROCESSING
DATA FUSION
SCALE
ACTION
PRODUCT +
QUANTITY
USER
FINAL OUTPUT
CARTS
TRIGGER
For each user we compute
the trajectories of the
distances between relevant
joints and the scale, around
the timestamp of the action.
We train the model to classify
the action on this data.
Assignment problem
wrist
elbow
shoulder
We defined some metrics to evaluate how well the
system is performing:
Metrics
RECORD DATA
ANNOTATION TOOL
CALCULATE,
STORE AND
ANALYZE METRICS
The same metrics can be defined in spaces where we
ignore either the user or the action variable.
We also evaluate these metrics on the space of the carts.
Dataset for Automated Store
Annotation tool in collaboration with https://itrexgroup.com
Synthetic 3D store rendering
Architecture
Camera-0
Camera-n
... VideoCapture
- Triangulation
- Track creation
VideoManager
- Pose estimation
TensorRT
InferenceServer
Scale-0
Scale-9
...
Scale-0
Scale-9
...
Gateway-0
Gateway-n
... - Pick/Drop classification
- Product classification
- User assignment
DataFusion
- Visualization
- Config change
Dashboard
- Carts update
- Config management
- Check-in/out handling
Backend
- Authentication
- Check-in
CheckInUI
- Payment
- Check-out
CheckOutUI
MongoDB
MQTT p/s
Network
CANbus
dbactions
Content
Frontend/Backend
● Visualize what’s happening
○ Users & carts
○ 3D reconstructions
● High-level visualization of
system’s status
○ Scales gateways status
○ connection errors
● Interface for store
configuration
○ Racks & scales spatial layout
○ Planograms
○ Products DB
○ Cameras configurations
Toolbar
App
Snackbar
Cameras
Carts
Products
Racks
Shelf
Tracking3D
MQTT
Checkin / Checkout frontend
● AngularJS webapps for checkin and
checkout UI
● Customer interaction
● Show information and feedback to
the customer

More Related Content

Similar to How to fully automate a store.pptx

Low-cost data-driven 3D reconstruction and its applications @ 6th ICE 3D Body...
Low-cost data-driven 3D reconstruction and its applications @ 6th ICE 3D Body...Low-cost data-driven 3D reconstruction and its applications @ 6th ICE 3D Body...
Low-cost data-driven 3D reconstruction and its applications @ 6th ICE 3D Body...
Alfredo BALLESTER FERNÁNDEZ
 
Thesis Ben O'Brien D11128055
Thesis Ben O'Brien D11128055Thesis Ben O'Brien D11128055
Thesis Ben O'Brien D11128055Ben O Brien
 
Architecting IoT with Machine Learning
Architecting IoT with Machine LearningArchitecting IoT with Machine Learning
Architecting IoT with Machine Learning
Rudradeb Mitra
 
Data Summer Conf 2018, “Architecting IoT system with Machine Learning (ENG)” ...
Data Summer Conf 2018, “Architecting IoT system with Machine Learning (ENG)” ...Data Summer Conf 2018, “Architecting IoT system with Machine Learning (ENG)” ...
Data Summer Conf 2018, “Architecting IoT system with Machine Learning (ENG)” ...
Provectus
 
Questions On The Equation For Regression
Questions On The Equation For RegressionQuestions On The Equation For Regression
Questions On The Equation For Regression
Tiffany Sandoval
 
IRJET - New Generation Multilevel based Atm Security System
IRJET - New Generation Multilevel based Atm Security SystemIRJET - New Generation Multilevel based Atm Security System
IRJET - New Generation Multilevel based Atm Security System
IRJET Journal
 
IRJET- Note to Coin Converter
IRJET- Note to Coin ConverterIRJET- Note to Coin Converter
IRJET- Note to Coin Converter
IRJET Journal
 
IRJET- New Generation Multilevel based Atm Security System
IRJET- New Generation Multilevel based Atm Security SystemIRJET- New Generation Multilevel based Atm Security System
IRJET- New Generation Multilevel based Atm Security System
IRJET Journal
 
Security by user through application to lock/unlock machine by face detection
Security by user through application to lock/unlock machine by face detectionSecurity by user through application to lock/unlock machine by face detection
Security by user through application to lock/unlock machine by face detection
IRJET Journal
 
Machine Learning for (DF)IR with Velociraptor: From Setting Expectations to a...
Machine Learning for (DF)IR with Velociraptor: From Setting Expectations to a...Machine Learning for (DF)IR with Velociraptor: From Setting Expectations to a...
Machine Learning for (DF)IR with Velociraptor: From Setting Expectations to a...
Chris Hammerschmidt
 
Smart Traffic Monitoring System Report
Smart Traffic Monitoring System ReportSmart Traffic Monitoring System Report
Smart Traffic Monitoring System Report
ALi Baker
 
IRJET- E-Gatepass System
IRJET- E-Gatepass SystemIRJET- E-Gatepass System
IRJET- E-Gatepass System
IRJET Journal
 
IRJET- Review on Colored Object Sorting System using Arduino UNO
IRJET- Review on Colored Object Sorting System using Arduino UNOIRJET- Review on Colored Object Sorting System using Arduino UNO
IRJET- Review on Colored Object Sorting System using Arduino UNO
IRJET Journal
 
ATM fraud detection system using machine learning algorithms
ATM fraud detection system using machine learning algorithmsATM fraud detection system using machine learning algorithms
ATM fraud detection system using machine learning algorithms
IRJET Journal
 
Partial Object Detection in Inclined Weather Conditions
Partial Object Detection in Inclined Weather ConditionsPartial Object Detection in Inclined Weather Conditions
Partial Object Detection in Inclined Weather Conditions
IRJET Journal
 
Design, Analysis and Fabrication of Pick & Place Colour Sorting Robotic Arm
Design, Analysis and Fabrication of Pick & Place Colour Sorting Robotic ArmDesign, Analysis and Fabrication of Pick & Place Colour Sorting Robotic Arm
Design, Analysis and Fabrication of Pick & Place Colour Sorting Robotic Arm
IRJET Journal
 
REAL-TIME OBJECT DETECTION USING OPEN COMPUTER VISION
REAL-TIME OBJECT DETECTION USING OPEN COMPUTER VISIONREAL-TIME OBJECT DETECTION USING OPEN COMPUTER VISION
REAL-TIME OBJECT DETECTION USING OPEN COMPUTER VISION
IRJET Journal
 
Flow Trajectory Approach for Human Action Recognition
Flow Trajectory Approach for Human Action RecognitionFlow Trajectory Approach for Human Action Recognition
Flow Trajectory Approach for Human Action Recognition
IRJET Journal
 
IRJET- Intrusion Detection through Image Processing and Getting Notified ...
IRJET-  	  Intrusion Detection through Image Processing and Getting Notified ...IRJET-  	  Intrusion Detection through Image Processing and Getting Notified ...
IRJET- Intrusion Detection through Image Processing and Getting Notified ...
IRJET Journal
 
IRJET - Cardless ATM
IRJET -  	  Cardless ATMIRJET -  	  Cardless ATM
IRJET - Cardless ATM
IRJET Journal
 

Similar to How to fully automate a store.pptx (20)

Low-cost data-driven 3D reconstruction and its applications @ 6th ICE 3D Body...
Low-cost data-driven 3D reconstruction and its applications @ 6th ICE 3D Body...Low-cost data-driven 3D reconstruction and its applications @ 6th ICE 3D Body...
Low-cost data-driven 3D reconstruction and its applications @ 6th ICE 3D Body...
 
Thesis Ben O'Brien D11128055
Thesis Ben O'Brien D11128055Thesis Ben O'Brien D11128055
Thesis Ben O'Brien D11128055
 
Architecting IoT with Machine Learning
Architecting IoT with Machine LearningArchitecting IoT with Machine Learning
Architecting IoT with Machine Learning
 
Data Summer Conf 2018, “Architecting IoT system with Machine Learning (ENG)” ...
Data Summer Conf 2018, “Architecting IoT system with Machine Learning (ENG)” ...Data Summer Conf 2018, “Architecting IoT system with Machine Learning (ENG)” ...
Data Summer Conf 2018, “Architecting IoT system with Machine Learning (ENG)” ...
 
Questions On The Equation For Regression
Questions On The Equation For RegressionQuestions On The Equation For Regression
Questions On The Equation For Regression
 
IRJET - New Generation Multilevel based Atm Security System
IRJET - New Generation Multilevel based Atm Security SystemIRJET - New Generation Multilevel based Atm Security System
IRJET - New Generation Multilevel based Atm Security System
 
IRJET- Note to Coin Converter
IRJET- Note to Coin ConverterIRJET- Note to Coin Converter
IRJET- Note to Coin Converter
 
IRJET- New Generation Multilevel based Atm Security System
IRJET- New Generation Multilevel based Atm Security SystemIRJET- New Generation Multilevel based Atm Security System
IRJET- New Generation Multilevel based Atm Security System
 
Security by user through application to lock/unlock machine by face detection
Security by user through application to lock/unlock machine by face detectionSecurity by user through application to lock/unlock machine by face detection
Security by user through application to lock/unlock machine by face detection
 
Machine Learning for (DF)IR with Velociraptor: From Setting Expectations to a...
Machine Learning for (DF)IR with Velociraptor: From Setting Expectations to a...Machine Learning for (DF)IR with Velociraptor: From Setting Expectations to a...
Machine Learning for (DF)IR with Velociraptor: From Setting Expectations to a...
 
Smart Traffic Monitoring System Report
Smart Traffic Monitoring System ReportSmart Traffic Monitoring System Report
Smart Traffic Monitoring System Report
 
IRJET- E-Gatepass System
IRJET- E-Gatepass SystemIRJET- E-Gatepass System
IRJET- E-Gatepass System
 
IRJET- Review on Colored Object Sorting System using Arduino UNO
IRJET- Review on Colored Object Sorting System using Arduino UNOIRJET- Review on Colored Object Sorting System using Arduino UNO
IRJET- Review on Colored Object Sorting System using Arduino UNO
 
ATM fraud detection system using machine learning algorithms
ATM fraud detection system using machine learning algorithmsATM fraud detection system using machine learning algorithms
ATM fraud detection system using machine learning algorithms
 
Partial Object Detection in Inclined Weather Conditions
Partial Object Detection in Inclined Weather ConditionsPartial Object Detection in Inclined Weather Conditions
Partial Object Detection in Inclined Weather Conditions
 
Design, Analysis and Fabrication of Pick & Place Colour Sorting Robotic Arm
Design, Analysis and Fabrication of Pick & Place Colour Sorting Robotic ArmDesign, Analysis and Fabrication of Pick & Place Colour Sorting Robotic Arm
Design, Analysis and Fabrication of Pick & Place Colour Sorting Robotic Arm
 
REAL-TIME OBJECT DETECTION USING OPEN COMPUTER VISION
REAL-TIME OBJECT DETECTION USING OPEN COMPUTER VISIONREAL-TIME OBJECT DETECTION USING OPEN COMPUTER VISION
REAL-TIME OBJECT DETECTION USING OPEN COMPUTER VISION
 
Flow Trajectory Approach for Human Action Recognition
Flow Trajectory Approach for Human Action RecognitionFlow Trajectory Approach for Human Action Recognition
Flow Trajectory Approach for Human Action Recognition
 
IRJET- Intrusion Detection through Image Processing and Getting Notified ...
IRJET-  	  Intrusion Detection through Image Processing and Getting Notified ...IRJET-  	  Intrusion Detection through Image Processing and Getting Notified ...
IRJET- Intrusion Detection through Image Processing and Getting Notified ...
 
IRJET - Cardless ATM
IRJET -  	  Cardless ATMIRJET -  	  Cardless ATM
IRJET - Cardless ATM
 

Recently uploaded

DevOps and Testing slides at DASA Connect
DevOps and Testing slides at DASA ConnectDevOps and Testing slides at DASA Connect
DevOps and Testing slides at DASA Connect
Kari Kakkonen
 
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...
DanBrown980551
 
PHP Frameworks: I want to break free (IPC Berlin 2024)
PHP Frameworks: I want to break free (IPC Berlin 2024)PHP Frameworks: I want to break free (IPC Berlin 2024)
PHP Frameworks: I want to break free (IPC Berlin 2024)
Ralf Eggert
 
Key Trends Shaping the Future of Infrastructure.pdf
Key Trends Shaping the Future of Infrastructure.pdfKey Trends Shaping the Future of Infrastructure.pdf
Key Trends Shaping the Future of Infrastructure.pdf
Cheryl Hung
 
A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...
sonjaschweigert1
 
UiPath Test Automation using UiPath Test Suite series, part 4
UiPath Test Automation using UiPath Test Suite series, part 4UiPath Test Automation using UiPath Test Suite series, part 4
UiPath Test Automation using UiPath Test Suite series, part 4
DianaGray10
 
PCI PIN Basics Webinar from the Controlcase Team
PCI PIN Basics Webinar from the Controlcase TeamPCI PIN Basics Webinar from the Controlcase Team
PCI PIN Basics Webinar from the Controlcase Team
ControlCase
 
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
Sri Ambati
 
Welocme to ViralQR, your best QR code generator.
Welocme to ViralQR, your best QR code generator.Welocme to ViralQR, your best QR code generator.
Welocme to ViralQR, your best QR code generator.
ViralQR
 
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdfObservability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Paige Cruz
 
By Design, not by Accident - Agile Venture Bolzano 2024
By Design, not by Accident - Agile Venture Bolzano 2024By Design, not by Accident - Agile Venture Bolzano 2024
By Design, not by Accident - Agile Venture Bolzano 2024
Pierluigi Pugliese
 
FIDO Alliance Osaka Seminar: Passkeys at Amazon.pdf
FIDO Alliance Osaka Seminar: Passkeys at Amazon.pdfFIDO Alliance Osaka Seminar: Passkeys at Amazon.pdf
FIDO Alliance Osaka Seminar: Passkeys at Amazon.pdf
FIDO Alliance
 
Monitoring Java Application Security with JDK Tools and JFR Events
Monitoring Java Application Security with JDK Tools and JFR EventsMonitoring Java Application Security with JDK Tools and JFR Events
Monitoring Java Application Security with JDK Tools and JFR Events
Ana-Maria Mihalceanu
 
Assuring Contact Center Experiences for Your Customers With ThousandEyes
Assuring Contact Center Experiences for Your Customers With ThousandEyesAssuring Contact Center Experiences for Your Customers With ThousandEyes
Assuring Contact Center Experiences for Your Customers With ThousandEyes
ThousandEyes
 
Essentials of Automations: Optimizing FME Workflows with Parameters
Essentials of Automations: Optimizing FME Workflows with ParametersEssentials of Automations: Optimizing FME Workflows with Parameters
Essentials of Automations: Optimizing FME Workflows with Parameters
Safe Software
 
State of ICS and IoT Cyber Threat Landscape Report 2024 preview
State of ICS and IoT Cyber Threat Landscape Report 2024 previewState of ICS and IoT Cyber Threat Landscape Report 2024 preview
State of ICS and IoT Cyber Threat Landscape Report 2024 preview
Prayukth K V
 
Introduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - CybersecurityIntroduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - Cybersecurity
mikeeftimakis1
 
Elevating Tactical DDD Patterns Through Object Calisthenics
Elevating Tactical DDD Patterns Through Object CalisthenicsElevating Tactical DDD Patterns Through Object Calisthenics
Elevating Tactical DDD Patterns Through Object Calisthenics
Dorra BARTAGUIZ
 
Transcript: Selling digital books in 2024: Insights from industry leaders - T...
Transcript: Selling digital books in 2024: Insights from industry leaders - T...Transcript: Selling digital books in 2024: Insights from industry leaders - T...
Transcript: Selling digital books in 2024: Insights from industry leaders - T...
BookNet Canada
 
Bits & Pixels using AI for Good.........
Bits & Pixels using AI for Good.........Bits & Pixels using AI for Good.........
Bits & Pixels using AI for Good.........
Alison B. Lowndes
 

Recently uploaded (20)

DevOps and Testing slides at DASA Connect
DevOps and Testing slides at DASA ConnectDevOps and Testing slides at DASA Connect
DevOps and Testing slides at DASA Connect
 
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...
 
PHP Frameworks: I want to break free (IPC Berlin 2024)
PHP Frameworks: I want to break free (IPC Berlin 2024)PHP Frameworks: I want to break free (IPC Berlin 2024)
PHP Frameworks: I want to break free (IPC Berlin 2024)
 
Key Trends Shaping the Future of Infrastructure.pdf
Key Trends Shaping the Future of Infrastructure.pdfKey Trends Shaping the Future of Infrastructure.pdf
Key Trends Shaping the Future of Infrastructure.pdf
 
A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...
 
UiPath Test Automation using UiPath Test Suite series, part 4
UiPath Test Automation using UiPath Test Suite series, part 4UiPath Test Automation using UiPath Test Suite series, part 4
UiPath Test Automation using UiPath Test Suite series, part 4
 
PCI PIN Basics Webinar from the Controlcase Team
PCI PIN Basics Webinar from the Controlcase TeamPCI PIN Basics Webinar from the Controlcase Team
PCI PIN Basics Webinar from the Controlcase Team
 
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
 
Welocme to ViralQR, your best QR code generator.
Welocme to ViralQR, your best QR code generator.Welocme to ViralQR, your best QR code generator.
Welocme to ViralQR, your best QR code generator.
 
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdfObservability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
 
By Design, not by Accident - Agile Venture Bolzano 2024
By Design, not by Accident - Agile Venture Bolzano 2024By Design, not by Accident - Agile Venture Bolzano 2024
By Design, not by Accident - Agile Venture Bolzano 2024
 
FIDO Alliance Osaka Seminar: Passkeys at Amazon.pdf
FIDO Alliance Osaka Seminar: Passkeys at Amazon.pdfFIDO Alliance Osaka Seminar: Passkeys at Amazon.pdf
FIDO Alliance Osaka Seminar: Passkeys at Amazon.pdf
 
Monitoring Java Application Security with JDK Tools and JFR Events
Monitoring Java Application Security with JDK Tools and JFR EventsMonitoring Java Application Security with JDK Tools and JFR Events
Monitoring Java Application Security with JDK Tools and JFR Events
 
Assuring Contact Center Experiences for Your Customers With ThousandEyes
Assuring Contact Center Experiences for Your Customers With ThousandEyesAssuring Contact Center Experiences for Your Customers With ThousandEyes
Assuring Contact Center Experiences for Your Customers With ThousandEyes
 
Essentials of Automations: Optimizing FME Workflows with Parameters
Essentials of Automations: Optimizing FME Workflows with ParametersEssentials of Automations: Optimizing FME Workflows with Parameters
Essentials of Automations: Optimizing FME Workflows with Parameters
 
State of ICS and IoT Cyber Threat Landscape Report 2024 preview
State of ICS and IoT Cyber Threat Landscape Report 2024 previewState of ICS and IoT Cyber Threat Landscape Report 2024 preview
State of ICS and IoT Cyber Threat Landscape Report 2024 preview
 
Introduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - CybersecurityIntroduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - Cybersecurity
 
Elevating Tactical DDD Patterns Through Object Calisthenics
Elevating Tactical DDD Patterns Through Object CalisthenicsElevating Tactical DDD Patterns Through Object Calisthenics
Elevating Tactical DDD Patterns Through Object Calisthenics
 
Transcript: Selling digital books in 2024: Insights from industry leaders - T...
Transcript: Selling digital books in 2024: Insights from industry leaders - T...Transcript: Selling digital books in 2024: Insights from industry leaders - T...
Transcript: Selling digital books in 2024: Insights from industry leaders - T...
 
Bits & Pixels using AI for Good.........
Bits & Pixels using AI for Good.........Bits & Pixels using AI for Good.........
Bits & Pixels using AI for Good.........
 

How to fully automate a store.pptx

  • 1. How to fully automate a store A Made in Italy use case
  • 3. Alessio Elmi Artificial Intelligence Engineer Shruti Verma Artificial Intelligence Engineer Michele Toni Artificial Intelligence Engineer Bruno Abbate Machine Learning Engineer linkedin.com/in/alessioelm i linkedin.com/in/shrutiverma2 linkedin.com/in/bruno-abbate linkedin.com/in/michele-toni Naser Derakhshan Computer Scientist linkedin.com/in/naser-derakhshan-51951828 Pietro Tortella Mathematician linkedin.com/in/pietro-tortella-976839ab Luca Lulleri Industrial Designer linkedin.com/in/lucalulleri Alessandro Re Machine Learning Engineer linkedin.com/in/akiross Riccardo Di Guida Machine Learning Engineer linkedin.com/in/riccardo-di-guida- 005764124 Davide Mazzini Deep Learning Engineer linkedin.com/in/davidemazzini Mattia Santachiara AR/VR Engineer linkedin.com/in/mattia-santachiara-90a1b379 Igor Moiseev Crazy CTO linkedin.com/in/moiseevigor 4 PhD 8 MSc
  • 5. Three main ML problems Object tracking and Anomaly detection Pose-estimation and People Tracking Assignment problem
  • 6. It was required a system which could validate the correct amount of goods picked up or dropped at the same time from a user. Hardware Design
  • 7. Scales PCB/Firmware The PICK action (bottle of water) The DROP action (bottle of water)
  • 8. Camera Positioning Study 1. Retrieve Cad Drawing of the space 2. 3D modeling of the space 3. Define camera position and direction 4. Grasshopper algorithm to make
  • 9.
  • 10. Camera Positioning Study Evolutionary Algorithm and Particle Swarm Optimization to optimize camera positioning.
  • 13. Pose-estimation ● Train 2D pose estimation model using a top view dataset including renderings from the synthetic datasets ● GPU version of the upsampling model (main bottleneck right now) ● Cameras “software” synchronization ● Reduce CPU and GPU load
  • 15. Tracking: The problem Match 2D-pose detections from different cameras to create 3D-tracks.
  • 16. Tracking: The glossary Detection One pose in a given frame at a given time Reconstruction Many detections different frames at a given time Track Many reconstructions at different times
  • 17. Tracking: The approach Hypergraphs for Joint Multi-view Reconstruction and Multi-object Tracking by M. Hofmann, D. Wold, G. Rigoll. 2013
  • 18. Tracking: The approach ● Construct all possible reconstructions and links ● Associate probabilities to them ● Associate probabilities to links ● Create Hypergraph ● Reduce to BIP problem ● Boolean variable per vertex ● Boolean variable per link ● Two constraints per vertex ○ Incoming flow = vertex variable value ○ Outgoing flow = vertex variable value ● Additional constraints from detections ○ Each detection might belong to at most one flow ● Cost per vertex variable from reconstruction prob ● Cost per link variable from link prob ● Minimize cost of flow Binary Integer Programming Minimize cost with integer variables satisfying given constraints
  • 19. Tracking: The approach online Window 0 Window 2 Window 1 Window 3
  • 20. ● Stabletracks ● Flexible ● No ID switch ● CPU Expensive ● Complexity ● Sensible to parameters calibration ● BIP is NP-Hard
  • 21. Tracking: The doing Introduced the 3D geometry of the store. ● Use geometric informations on cameras and obstacles to filter reconstructions ● Make all parameters position-dependent RESULTS: ➔ Lighter graph (-50% variables, -20% equations) ➔ Reduced complexity → Better scalability to bigger stores
  • 22.
  • 23. Object tracking and Anomaly detection
  • 24. � Detect misplaced products in the scales � Detect extraneous objects in the scales Object tracking and Anomaly detection
  • 25. 1) ODIN (Out-of-distribution detection) 2) Reconstruction-based using Autoencoder 3) Object Detection using ResNet 50 + Faster RCNN Object tracking and Anomaly detection
  • 27.
  • 29. Assignment problem The aim is to combine data from cameras and scales to predict events e = (timestamp, action, scale, product, quantity, user) 2 INPUT SOURCES CAMERAS SCALES DATA PROCESSING DATA PROCESSING DATA FUSION SCALE ACTION PRODUCT + QUANTITY USER FINAL OUTPUT CARTS TRIGGER
  • 30. For each user we compute the trajectories of the distances between relevant joints and the scale, around the timestamp of the action. We train the model to classify the action on this data. Assignment problem wrist elbow shoulder
  • 31. We defined some metrics to evaluate how well the system is performing: Metrics RECORD DATA ANNOTATION TOOL CALCULATE, STORE AND ANALYZE METRICS The same metrics can be defined in spaces where we ignore either the user or the action variable. We also evaluate these metrics on the space of the carts.
  • 33. Annotation tool in collaboration with https://itrexgroup.com
  • 34.
  • 35.
  • 36. Synthetic 3D store rendering
  • 38. Camera-0 Camera-n ... VideoCapture - Triangulation - Track creation VideoManager - Pose estimation TensorRT InferenceServer Scale-0 Scale-9 ... Scale-0 Scale-9 ... Gateway-0 Gateway-n ... - Pick/Drop classification - Product classification - User assignment DataFusion - Visualization - Config change Dashboard - Carts update - Config management - Check-in/out handling Backend - Authentication - Check-in CheckInUI - Payment - Check-out CheckOutUI MongoDB MQTT p/s Network CANbus dbactions
  • 39. Content Frontend/Backend ● Visualize what’s happening ○ Users & carts ○ 3D reconstructions ● High-level visualization of system’s status ○ Scales gateways status ○ connection errors ● Interface for store configuration ○ Racks & scales spatial layout ○ Planograms ○ Products DB ○ Cameras configurations Toolbar App Snackbar Cameras Carts Products Racks Shelf Tracking3D MQTT
  • 40. Checkin / Checkout frontend ● AngularJS webapps for checkin and checkout UI ● Customer interaction ● Show information and feedback to the customer

Editor's Notes

  1. https://arvrjourney.com/human-pose-estimation-using-openpose-with-tensorflow-part-2-e78ab9104fc8
  2. Takes into account: Back-projection False positive rate False negative rate Expected detections of the scene
  3. m1 tells us how many events are well predicted among all these possible outcomes. m2 is the index of how many ground truth events were well predicted. m3 tells us how many predictions were right.