Documentation requirements

System communication

  • FONCT01 : Our system should be able to broadcast an audio alert message to employees during a danger using a loudspeaker.

  • FONCT02 : Our system must be able to communicate with different areas of the site via an API.

  • FONCT03 : Our system should be able to record audio messages via microphones.

  • FONCT08 : Our system should be able to contact assistance graduated according to the severity, which it will define according to the situation

  • FONCT09 : Our system should be able to contact surrounding employees using loudspeakers.

  • FONCT10 : Our system should be able to contact the site security managers using telephone calls via VOIP or a call server and via emails via an SMTP server

  • FONCT11 : Our system should be able to make outside calls via VOIP or a call server in order to contact the emergency services in the event of a serious danger.

  • FONCT13 : Our system must be able to communicate its analysis data with other modeling software (RAD) via modules and API

  • CONST01 : Our system must respect the constraints linked to video recording and broadcasting.

Storage

  • FONCT04 : Our system must be able to record the data collected on a database using Postgres SQL DBMS.

  • FONCT05 : Our system must be able to record video or audio data on a storage space such as QNAP TS-453D NAS

Input data

  • FONCT07 : Our system must be able to capture images via a 10 GigE machine vision camera type camera: ORYX range | FLIR SYSTEMS, on which avi / mp4 format video will be recorded and processed.

  • CONST02 : Our system must respect the constraints related to audio recording and broadcasting.

Data processing

  • FONCT06 : Our system must be able to detect when a person falls through the use of a Computer Vision solution that analyzes the kinematics of the body.

  • FONCT12 : Our system must be able to provide an accurate description of a movement using the ragdoll wireframe of a video image

  • FONCT15 : Our system should be able to provide a map of the risk areas of a site, a summary dashboard, a report on the dangerousness of an area via a web interface that will communicate with an API.

  • FONCT16 : Our system should be able to provide meaningful analytical data to the medical / ergonomist profession, notably based on the cause tree method which allows the analysis of a work accident.

  • FONCT17 : Our system should be able to anticipate the dangerousness from the statistics obtained.

  • FONCT18 : The system must be able to offer a representative analysis of the movements made and analyze the movements of several people.

  • NOFONCT19 : The vision computing system will use the OpenCV language and libraries for image processing, TensorFlow-gpu for situation processing.

Hardware

  • NOFONCT14 : Our system could be able to operate autonomously via a current source whose overall power consumption is limited to 5W up to 20W at peak load and be powered by a 20Wh battery.

  • NOFONCT20 : The system will use a GPU compatible with Nvidia / AMD drivers to be Cuda / Rocm compatible

Devops

  • NOFONCT21 : The vision computing system will be separated into a micro service orchestrated by Kubernate and docked via Docker.

  • NOFONCT22 : The development environment will use the development pipelines that will integrate the KuberFlow tool, and a development map will be used.