What is eyeAI?

EyeAI is a software package for indoor analytics. With eyAI you can understand your visitors, behavior and preferences better and shape your service/product accordingly. eyeAI provides insights on:

Frequent
visitors tracking
Detection of popular and ignored areas within the indoor area
Visitors’
trajectories sampling
Timeline of visitors activity displayed in top-down view
Number
of visitors
Number of engaged visitors (e.g. shoppers)
Time of entry
and exit

With eyeAI you can improve your visitors’ understanding and shape your services or product along with this information.

When reviewing the current state of indoor analytics, we found out that most of solutions come with beacons, mobile phones tracking and other sources that rely either on custom hardware or customer engagement with local wifi connection. The advantage of eyeAI is that all it needs is the information about the installed security cameras and videos from the cameras. With eyeAI, indoor analytics is significantly simplified, and is more cost effective.

Our tech processes videos from cameras and renders stats in a 2D top-down view. So far we need at least two cameras with an overlapping field of view, but we plan to experimentally enable a single-camera operation mode in 2019.

eyeAI cares about visitors’ confidentiality, that’s why there is no personal data exposed.

Techstack and Architecture

Location detection and tracking microservices are realized as docker containers exposing REST interface.

Frontend:
Angular
Backend:
Python, Flask
Analysis:
Python, Tensorflow, OpenCV
Data generation:
Python, Blender

Vizualization

Roadmap

2018 Q1-Q2

  • Initial testing of face detection and recognition models
  • Testing of human detection model.
  • Implementation of object tracking

2018 Q3

  • Prototype backend for customer location detection (working with 2 cameras)
  • Prototype backend and UI for analysis of frequent customers appearances

2018 Q4

  • Prototype backend and UI working with 2+ cameras
  • Landing page
  • Formulating acceptable camera locations and orientations
  • Data collection for customer location accuracy measurements

2019 Q1

  • Pipeline optimization, preparing code and infrastructure for production level
  • Polishing UI/UX
  • Rolling out system at selected locations, collecting data and feedback

2019 Q2

  • Collecting data and training POC model for estimation of customer location from a single camera, i.e. simplified scene reconstruction
  • Adding automatic detection of customer profile: favourite areas, emotions, preferred colors etc

Contact us

  • We're here to give you more information about eyeAI
    and answer any questions you may have.
  • info@eye-ai.tech
Top