/InfosysHack

Submission for ML/AI challenge in Infosys Digital Makeathon 2018 on the use case "ATM Surveillance in Banks" which fetched us the runners up position. System can perform face recognition, weapon and potential harmful object detection.

Primary LanguageJupyter Notebook

InfyHack

This repository contains the code to the features incorporated in ML/AI challenge of Infosys Digital Makeathon'18, on the use case "ATM Surveillance in Banks" which fetched us the runners up position.

The functions perfomed by our ML/AI model are:

  1. Face Recognition - Using FaceNet and haar cascades
  2. Weapon and Skimming Device Detection - using Fast-RCNN
  3. Helmet and Sunglass Detection - using CNN and haar cascades

An extension to this will be anomalous activity tracking using LSTM and CNN.

Demo

Alt text