Online feature extraction
and event generation for
computer-animal
interaction
Leon Bonde Larsen & Mathias Neerup
Advanced Computer Systems Group
Introduction
Introduction
ATS v2.0
Problem: Stateless HTTP protocol
State using cookies
State using cookies
Memory
Server process
Process
Server side playing
somePlayer
Server side playing > Multiple users
Server side playing > Multiple users > Containers
Streams
Streams > Producers
Streams > Consumers
Streams > Historian
Streams > Processors
Event Database
Streams > Implementation
ATS v1.0
ATS v1.0> Limitations
▪ Janus WebRTC
▪ One user at a time
ATS v2.0
▪ Docker swarm
▪ React
▪ Redux
▪ SocketIO
▪ Semantic UI
ATS v2.0 > Janus
Janus
- Janus
- Input: Stream
ATS v2.0 > Backend
Backend
- Player
- Avahi Info.
ATS v2.0 > Plugin
Plugin
- A react
component
- Inputs
- Outputs
- Image
ATS v2.0 > Services
ATS v2.0 > Services
ATS v2.0 > Services
ATS v2.0 > Services
Container (service) pooling
▪ Pooler
▪ Service
▪ Acquired
▪ Released
▪ Policies
▪ Memory
Container (service) pooling
▪ Label on service
▪ SessionID
▪ Release policy
▪ 24 hours
▪ Closing session
Multicast traffic
▪ Multicast streams
▪ Enable IPv6
▪ Connect to bridge
https://github.com/mcproxy/mcproxy
Multicast traffic
▪ Multicast streams
▪ Avahi
▪ Avahi-daemon
▪ reflector
https://github.com/mcproxy/mcproxy
CD/CI
▪ Gitlab docker registry
CD/CI - imageUpdater
▪ Gitlab docker registry
CD/CI - deploy strategy
▪ Gitlab docker registry
Software stack
Hardware
Docker swarm is suitable in biological research
Summary
Demo
Questions ?
Leon Bonde Larsen Mads Tilgaard Jensen Mathias Neerup
Ph.d Student programmer Research Assistant
https://acsg.sdu.dk
Demo - Video
Results

DCSF 19 Online Feature Extraction and Event Generation for Computer-Animal Interaction