/HackMx-TeamTuring

CrowdLytics is Google Analytics, for the retail store. | Second Place at HackMX2 Google's Challenge

Primary LanguagePython

CrowdLytics

CrowdLytics is a user experience analytics reporting tool that works on top of already existing security cameras. It uses a custom trained neural network to recognize faces and processes them using Google Vision API, getting store time, basic demographics, mood and other useful information to decision makers, marketing teams and managers. The information that’s gathered is completely anonymous, and cannot be linked to a specific person in any way once processed.

For Recruiters

  • Developed and trained a custom TensorFlow neural network to capture and recognize human faces from a live video with an accuracy rate of 53%.

  • Utilized the Google Cloud Platform Vision API to analyze human faces in search of meaningful demographics. Unfortunately, due to hardware limitations at the hackathon, no reliable results could be gathered.

  • Employed Firebase and Firestore to store the program’s data, enabling Real Time Analytics to be accesed from anywhere.

Screenshots

A windows explorer window containing multiple files that depict human faces.
Figure 1. Data Gathered
A picture depicting a Firebase storage instance on the left
Figure 2. Google Storage and Firebase
A rough sketch of a material design web interface
Figure 3. User Interface Prototype

Technologies

  • Python

  • Flask

  • TensorFlow

  • Google Cloud’s Platform Vision API

  • Google Cloud’s Platform Firebase

  • Google Cloud’s Platform Firestore

Contributors