Back to R&D main

Automated sheep counting for the live export industry

Did you know a new system being developed has the potential to help the export industry improve the efficiency of counting sheep?

Project start date: 03 July 2018
Project end date: 02 March 2020
Publication date: 06 April 2021
Project status: Completed
Livestock species: Sheep
Relevant regions: National, International

Summary

Keeping count of sheep being loaded onto a live export ship is currently a highly skilled and stressful task, as tens of thousands of sheep are typically carried on each voyage and loaded and unloaded continuously over 1-2 days. There are strict regulations around the number of animals allowed on a livestock export ship, so accuracy is vital. However, there is currently the potential for miscounts due to human error.

This project sought to build on systems using artificial intelligence in other industries to create a video-based automated sheep counting system, with a camera mounted over the loading ramp into ships.

The system was has the potential to free up staff at the port for other tasks while reducing the potential for discrepancies in the count between the start and end of a voyage.

Objectives

The objectives of this project were to:

  • develop a video camera-based algorithm system that demonstrates a level of at least 99.5% accuracy
  • complete a live demonstration
  • provide details and instructions to exporters on appropriate hardware, the software design and relevant camera setup.

Key findings

  • Accuracy was shown to reach 99.9% and above in various real-life scenarios, including at a shipping port and sale yards. 
  • Workers and dogs moving in and out of the video counting area were successfully excluded from the count.
  • Many saleyard managers have shown great interest in commercialising the technology for sheep counting and several demonstrations have been organised at saleyards.

Benefits to industry

A highly accurate automated system will provide independent calculations of the number of sheep being loaded onto export ships and reduce the human resources required to count sheep.

This technology is also likely to be suitable for use in feedlots and saleyards, and eventually for counting other types of livestock.

MLA action

MLA are taking steps to identify a commercial company to commercialise the sheep counting technology for the live export industry.

Future research

There will be more technologies for animal welfare and other applications, such as:

  • sheep behaviour analysis (lying/standing)
  • density control in pens (early warning to check the space occupancy)
  • animal body temperature estimation
  • weight estimation after the shearing 
  • cattle skeletal detection
  • cow facial recognition.

More information

Contact email: reports@mla.com.au
Primary researcher: University of Technology Sydney