Attenborough is a software suite being developed by Greenroom Robotics that uses a combination of computer vision and machine learning techniques to identify, classify and geo-tag animals (or anything with a heat signature).
The purpose of this is to provide in-depth statistics on the whereabouts and count of animal herds for land management. Invasive and damaging pests such as foxes may also be identified to provide a means of tracking and deploying countermeasures to ensure the health of the animals on the property.
Attenborough in its current form is being applied for post-processing UAV (Unmanned Aerial Vehicle) survey missions. Future developments will enable Attenborough to run in real-time during flight providing on the spot analysis and datasets to an intuitive user interface. This allowing the user to view, interrogate or add to the autonomous analysis on the fly or after the fact.
Attenborough already analyses footage faster than any human and with higher accuracy thus enabling tactical awareness or strategic data-driven insights. Decrease in processing time and increase in accuracy allow for the ability to survey larger areas of land with less delay and increased confidence.
Machine learning (ML) models associated with AI require large datasets and effort to ‘train’ to become artificially intelligent. We will work closely with our clients to speed up this whole process using our in-house computer vision collection and processing software. Over time and with more data the animal identification model will become larger enabling higher confidence and identification on a wider array of animal species to be identified.
Attenborough is still under development so that we can meet our goal of integrating cutting-edge computer vision with big data management and visualisation techniques for strategic land oversight.