Skip to content

Home

Jump to bottom
Gin edited this page Apr 29, 2021 · 5 revisions

Overview

Welcome to the 2021 FSI Autism Hackathon. This use case will explore the use of Azure Kinect/Webcam as an assistive technology during a therapy session for the identification of self-stimulatory or repetitive behaviors commonly associated with autism. The key goal is to aid with data collection of these repetitive behaviors and allow the therapist to focus more on the session vs. data collection. A secondary objective could be to aid in identification of autism in recognizing specific repetitive motions and behaviors that are indicative of autism, finding similarities in patterns and potential recommendations on treatments.

For this challenge, we will be partnering with LinedanceAI on this initiative. They created AI algorithms for human activities & movement recognition - the AVAR API runs over video or video streams to identify and analyze humans pattern of behaviors.

Suggested functionalities to build into the use case architecture:

  • A database for therapist, patient and session recordings with a simple GUI for management and appropriate security guardrails
  • A marketplace or repository of benchmark (pattern recognition) files that is shareable to different users
  • A metadata store for the time-series pattern recognition against therapy sessions
  • A GUI/visuals that matches the metadata to the session recording

Hackathon Objective

  • Based on the scenario and suggested functionalities, create a baseline architecture that would enable those functionalities.
  • Hack away on areas or functionalities that you are interested in developing.
  • Stitch together these modular components and package the services so they can be easily deployed or reused by others.

Open-source Objective

  • Create a platform that is scalable and reusable, reducing the barrier to entry for research or products to be built for this field.

LineDanceAI Endpoint and usage:

Clone this wiki locally