Drone Imagery Services logo
    The River Cocker at Cockermouth from the air, the riverbank flown for the invasive plant demonstration survey

    Invasive plant detection, recording and mitigation

    Invasive plant surveys by drone

    Find invasive plants from the air. Record where they are. Prove it later.

    We fly a site, and our AI-assisted models identify and locate Japanese knotweed, Himalayan balsam, giant hogweed and other invasive plants in the imagery. You get a dated, georeferenced record you can act on, monitor against, and rely on if a dispute ever arises.

    Invasive plant identification and location from drone imagery

    Invasive non-native plants are still found the slow way: someone walks the bank, the boundary or the lineside and writes down what they see. Across a catchment, a railway route or a council estate that means most of the ground is never looked at, and infestations are found late, when treatment costs most and spread has already happened.

    We are developing software that does the looking from the air. A drone flight captures the whole site at high resolution. AI models trained to recognise invasive plants review every frame and flag what they find, with a confidence level for each identification. The flagged plants are then located on a map, with the source photographs attached, so a land manager can see exactly what was found, where, and when. Our aim is simple: to be the UK's most useful provider of invasive plant records from the air, with a method that is open about its confidence and its limits.

    As a CAA-registered and fully insured operator we handle the whole job, from flight planning to the finished record. The same AI-assisted approach powers our survey data analysis and building observation reports.

    What it looks like in practice: River Cocker, Cockermouth

    In October 2026 we flew two stretches of the River Cocker as a demonstration survey for a Cumbrian rivers trust, to see what the models could find on a real riverbank at the end of the growing season. These are the figures from that record.

    435
    Photographs
    3,713
    Identifications
    929
    Located plants
    55
    Mosaic features
    Nadir orthomosaic of the River Cocker north of the A66 bridge, 7 October 2026, with located plants marked
    7 October 2026, about 300 m of the River Cocker north of the A66 road bridge. DJI Matrice 4E. Demonstration survey, not a commissioned one.
    • Himalayan balsam identified at high confidence in the low-level frames near the road bridge on 7 October, and widespread along the low-level stretch on 8 October, almost continuous along the near bank and cobble bars.
    • Every identification carries a confidence level (high, probable, possible) and links back to the frame it came from, so a surveyor can check it in seconds.
    • Four bushes flagged from 60 m were checked in lower frames and ruled out. The record shows the check, not just the answer.
    • No Japanese knotweed, giant hogweed, rhododendron, cherry laurel or snowberry seen on either day. A clean result is recorded as carefully as a positive one.
    • Balsam is only assessable in the low-level frames. The rest of the reach is marked unassessed, not clear. We say what we could not see.

    The full record, with photographs, mosaics, maps and the per-plant table, is held as a private client page. If you would like to see it, ask us and we will share access.

    How It Works

    1
    Fly

    We plan the flight for the plant and the season: low passes for identification, a nadir mosaic for mapping, repeat visits where monitoring is needed. All flown under our CAA registration and insurance.

    2
    Identify

    AI models review every photograph and flag candidate plants, each with a confidence level. Our own software runs the models and builds the record; a person checks what matters.

    3
    Locate and map

    Flagged plants are placed on the orthomosaic and on a map with grid references, so you can see extent, distance to boundaries, and change against earlier flights.

    4
    Record and report

    You receive a dated record: report, map layers, mosaic and photographs, with what was found, what was ruled out and what could not be assessed. Ready for a management plan or an evidence file.

    Species we look for

    We start with the plants that cost UK land managers the most. Detection confidence varies by species, season and flight height, and we say so on every record. Reference photos are shown for species we have not yet photographed ourselves and are credited below each card.

    Japanese knotweed stand seen from the air

    Reference photo: Alan Hughes, CC BY-SA 2.0, via Wikimedia Commons (cropped)

    Japanese knotweed

    Reynoutria japonica

    Distinctive zigzag stems and heart-shaped leaves in leaf; dense straw-coloured canes in winter. Detectable from ordinary drone imagery in the growing season. The focus of our 2027 trial.

    Priority species
    Himalayan balsam on a riverbank

    Our capture, River Cocker, October 2026

    Himalayan balsam

    Impatiens glandulifera

    Pink flowers and dense stands along watercourses from July. Identified at high confidence in our River Cocker demonstration.

    Demonstrated
    Giant hogweed

    Reference photo: M J Richardson, CC BY-SA 2.0, via Wikimedia Commons (cropped)

    Giant hogweed

    Heracleum mantegazzianum

    Large white umbels and tall stems, a public safety hazard as well as an ecological one. Checked for on every riverbank flight.

    In development
    Rhododendron ponticum thicket

    Reference photo: Robert Linsdell, CC BY 2.0, via Wikimedia Commons (cropped)

    Rhododendron ponticum

    Rhododendron ponticum

    Evergreen thickets that shade out woodland floors. Visible year-round from the air, which makes it a good candidate for winter flights.

    In development
    American skunk cabbage in wet woodland

    Reference photo: sylvia duckworth, CC BY-SA 2.0, via Wikimedia Commons (cropped)

    American skunk cabbage

    Lysichiton americanus

    Large yellow spathes in spring in wet woodland and ditches. A growing problem in Cumbrian catchments.

    In development
    Mixed riverbank vegetation

    Reference photo: Kolforn, CC BY-SA 4.0, via Wikimedia Commons (cropped)

    Other species

    Cherry laurel, snowberry, buddleia, cotoneaster

    If a plant matters on your land, tell us. Models can be trained on new species where enough imagery exists.

    On request

    Who It's For

    Rivers trusts and catchment partnerships mapping balsam, knotweed and hogweed at catchment scale
    Local authorities recording invasive plants on their land as part of a management plan and to mitigate liability
    Rail, highways and utility operators with long boundaries and neighbours on the other side of them
    Housing associations and estate owners who need to know what is on the land before works start
    Developers and contractors screening a site perimeter before acquisition or before ground is disturbed
    Knotweed and invasive plant specialists who want a dated aerial baseline alongside their ground survey
    Ecologists and researchers who need repeatable, georeferenced imagery across seasons
    Anyone who has been asked to prove when a plant was, or was not, present on a piece of land

    Why recording matters

    Invasive plant disputes are decided on dates and records: when the plant was present, how far it had spread, and when treatment started. The cases below are public judgments. A dated, georeferenced aerial record is the kind of evidence they turned on.

    Court of Appeal, 2018

    Network Rail v Williams and Waistell

    Knotweed encroaching from the railway onto two homes in Maesteg was held to be an actionable nuisance. Network Rail's treatment record had a gap of several years. Damages and costs followed.

    County Court, 2023

    Pember v Transport for Wales

    A Pontypridd homeowner was awarded close to £10,000. Transport for Wales said afterwards that the case had turned on a single missing documentary record of treatment.

    Supreme Court, 2024

    Davies v Bridgend County Borough Council

    The council won because the evidence showed the knotweed had spread before its duty arose and that treatment began in 2018. Dates of knowledge and action decided the outcome.

    We are not lawyers and this is not legal advice. Where a record is intended for use in a dispute we work alongside your appointed specialist or expert, and we design the survey to their requirements.

    What You Receive

    A dated report with a per-plant table: species, confidence, grid reference, source frame
    Georeferenced orthomosaic of the site with detections overlaid
    Map layers (GeoTIFF, KMZ and GIS-ready formats) for your own systems
    Every source photograph, so any identification can be checked
    A clear statement of what was ruled out and what could not be assessed
    Change against earlier flights where repeat surveys are flown
    A private online record page for your team, like the one we built for the River Cocker demonstration
    Flight logs, operator details and insurance confirmation for your file

    Where we are, and where this is going

    We would rather show you the state of the method than overclaim it. This is the programme as it stands.

    1. Published research

      Detection from ordinary drone imagery is established

      Peer-reviewed studies in Wales (2011, aerial photography) and the United States (2024 and 2025, drone imagery) report high-accuracy detection of Japanese knotweed in the growing season using standard RGB cameras. No UK drone dataset has yet been built to validate this on British sites.

    2. October 2026

      First demonstration survey, River Cocker

      435 photographs, 3,713 identifications, 929 located plants across two flights for a Cumbrian rivers trust. Himalayan balsam identified at high confidence; no knotweed, hogweed or rhododendron found. Delivered as a private client record.

    3. Autumn and winter 2026

      Building the method with researchers and land managers

      We are in conversation with university researchers in invasive plant ecology and image processing, and with rivers trusts who hold decades of ground records, to design a field trial whose data is useful for research as well as for our clients.

    4. 2027 growing season

      UK field trial on sites with known infestations

      Repeat flights across the season on partner sites, checked against ground truth, to build the first labelled UK drone dataset for Japanese knotweed and to measure where detection works and where it fails: season, lookalike species, canopy cover and small stands.

    5. Ongoing

      Open about limits

      Every record states confidence, flight height, season and what was not assessable. We will publish what the trial finds, including the failures.

    Frequently Asked Questions

    Published research says yes, from ordinary RGB imagery during the growing season, and our models are built on that basis. We have not yet validated knotweed detection on UK sites against ground truth; that is the purpose of our 2027 trial. Until then we describe knotweed detection as in development and we will tell you the confidence of every identification.

    Most invasive plants are easiest to identify in leaf or flower, roughly May to September for knotweed and July to September for Himalayan balsam. Rhododendron is visible year-round. Winter flights have their own uses: leaf-off conditions make the ground visible under trees, and dead knotweed canes persist. We will advise on timing for the plant and the site.

    Standard flights give positions good to a few metres, which is enough for management and monitoring. Where distance to a boundary matters, for example in a dispute, we fly with RTK positioning or ground control to tighten that to centimetres, and say in the record which was used.

    Yes. The models do the looking; we review the high-value identifications and anything borderline, and we mark on the record what was checked. Every identification links to its source photograph so your own specialist can verify it.

    No, and it is not meant to. A PCA-accredited surveyor's report is what lenders and insurers rely on for a property. What we provide is the wide-area screening and the dated aerial record that a specialist cannot gather on foot, and we are happy to work alongside your surveyor or expert.

    The record is dated, georeferenced and traceable to source photographs and flight logs, which is what evidence needs to be. Whether and how it is used is a matter for your legal advisers and any appointed expert; we will design the survey to their requirements.

    We are based in Cumbria and work across the UK. Rivers trusts, councils and infrastructure owners in the North West, North East, Wales and the Borders are closest to home, but we travel for the right project.

    Areas We Cover

    Based in Cumbria, we travel across the UK for survey and inspection work. We regularly operate throughout Cumbria and the Lake District, across the North West and North East, and up to the Scottish border, and take on projects nationwide.

    CumbriaLake DistrictNorth West EnglandNorth East EnglandScottish BorderWalesUK-wide

    Know what is on your land before someone else tells you

    Tell us about the site and the plants you are worried about, and we will come back with a plan and a quote. We are CAA-registered, fully insured and cover the whole UK from a Cumbria base.