Chapter 2 Platform Overview

The Sticky Pi project contains several interconnected tools. A github organisation features the material (source code, data, CAD files, …) for each individual tool.

Here is a list of all the tools:

Name	Description
sticky-pi-device	The Hardware and software for the Sticky Pi cameras (i.e. device)
sticky-pi-android-harvester	An android app to harvest the data and interact with the sticky pis
sticky-pi-api	Server and client api and the docker services for the server webapps
sticky-pi-ml	Machine learning for the Sticky Pi project
sticky-pi-manuscript	Manuscipt material and experiments for the 2021 method publication
sticky-pi.github.io	Source code of this documentation

This is an overview of the main components of the platform. The details of each part will be explain further.

• Hardware
  • Sticky Pi devices – Takes pictures of traps and record environmental conditions
  • Data harvester – Manages devices and retrieve their data (in the field)
• Web server
  • Database – Stores image metadata as well as processing results, users…
  • S3 server – Store and serve image data and machine learning models
  • API – Parses client requests to put/get resources from the underlying database
  • Webapp – A frontend website to visualise/plot the images taken and processing results
  • Nginx server – Secures and routes connections between the above services and the outside world
  • Analysis pipeline – Pre-processes images in real time
• Machine Learning
  • Universal Insect Detector – Detects insect vs background, in all images
  • Siamese Insect Matcher – Tracks insect instances across imaged
  • Insect Tuboid Classifier – Predicts insect taxonomy from a tracked instance