Detecting the presence of invasive plant species: More quickly, cheaply and safely using AI and machine vision

Invasive Non-Native Species (INNS) are organisms introduced into areas outside their native region where they then threaten ecosystems. They are regarded as one of the top five threats to biodiversity worldwide (IPBES, 2019), as well as having significant economic impacts, with companies in various sectors such as transport and utilities spending considerable time and resources to identify and remove them. Current methods for identifying the presence of INNS rely on ecological surveys, which are time consuming and costly, especially within road and rail infrastructure. Keen AI, the UK Centre for Ecology and Hydrology (CEH) and Time-Lapse Systems are combining their expertise in Artificial Intelligence (AI), INNS and image collection to provide a faster and more efficient method of conducting surveys of this kind.

Keen AI has expertise in providing AI solutions to companies such as National Grid, helping to streamline their visual condition assessment process. CEH have a long track record of research on invasive species and are pioneering image recognition services for Japanese Knotweed with the conveyancing sector. Time-Lapse systems are experts in capturing imagery for specialist applications. Our complementary experience, skills and resources provide an opportunity to develop a novel AI platform for detecting the presence of invasive species.

Current solutions for surveying an area for INNS include sending ecologists to perform a manual survey, which is time-consuming and costly, or the manual review of photographs taken from high definition digital cameras attached to drones or planes. Using AI technology, our proposal would reduce the time it takes to conduct an ecological survey of this kind, producing cost and time savings for the customer, and providing location specific information to support decision-making and management actions.

Our project vision is to assess the feasibility of developing an AI platform for detecting the presence of invasive plant species within linear infrastructure. This innovation will provide a rapid, high quality vegetation survey methodology, which will result in cost and time savings for our customers, and result in an increased understanding of market requirements for an AI innovation of this type. The project will have four key objectives:

1. Collection of vegetation imagery of sufficient quality;
2. Training of AI algorithms to identify INNS in the image dataset;
3. Processing high volumes of images to locate INNS geospatially; and
4. Evaluation of the AI model performance.