Impact case study database

150 to 200 species of life on our planet become extinct every day¹. This rate of “biological annihilation” means a sixth mass extinction in Earth’s history is under way², with catastrophic consequences for those ecosystems, the world economy, and planet as a whole. Indeed, biodiversity loss and consequent ecosystem collapse are one of the 5 foremost dangers facing humanity³. There is a fundamental need to routinely monitor animal populations over much of the globe so that conservation strategies can be optimized with such information.

The critical units of tracking change in animal populations are Animal Distributions and Densities (ADDs). However, most ADD surveys are conducted manually, which is extremely labour-intensive, inherently slow and costly. Since project initiation (2015), we are now (12/2020) working with 10 leading conservation agencies on 4 continents to overcome this problem. To assess the impact our thermal drone and artificial intelligence animal detection system has had on tackling biodiversity loss, we invited our partner conservation agencies to provide feedback on what impact, if any, the system has had on their organisation, from individual project level to national/global strategy. Corroborating Sources (CS) 1 – 5 provide examples of the letters we received. Below we summarise the reported impacts, grouped into 6 themes commonly covered by the organisations. References to feedback from specific organisations are shown as [CS#], where # = organisation number.

Impact 1 – Up to factor 100x improvement in ADD survey efficiency: Our system provides a step-change in ADD survey efficiency through a combination of (i) drones being able to cover large areas much more quickly than traditional ground surveys, (ii) thermal cameras being able to easily pick out animal’s heat signatures, and (iii) machine learning being able to automate identification of animals in footage that would traditionally have been done by eye.

¹ World Wildlife Fund for Nature (WWF): https://wwf.panda.org/knowledge_hub/endangered_species/

² Ceballos et al., Biological annihilation via the ongoing sixth mass extinction signaled by vertebrate population losses and declines, Proceedings of the National Academy of Sciences, 2017, 114, 30

³ World Economic Forum Global Risk Report: https://www.weforum.org/reports/the-global-risks-report-2020

Dr Hudson, Head of Research, Durrell Wildlife Conservation Trust: “… the team then flew the system over key areas of marsh that have been inaccessible to previous surveys. In 3 x 20-minute flights, the drone system reproduced lemur densities estimated from previous surveys requiring 24 person-weeks of effort. Scaling-up this programme to cover the entire lake represents an approximately 100 times increase in survey efficiency. … The adoption of this research technology will revolutionise the scale and frequency with which DWCT and our partners in Madagascar, including the Madagascar National Parks Service, can conduct surveys.” [CS3]

Dr Spaan, Population Monitoring Coordinator of the Mexican Conservation NGO, ‘ConMonoMaya’: “… the drone allows for much faster coverage of an area than the [traditional] ground method does.” [CS4]

Impact 2 – Up to factor ~15x improvement in processing ADD survey data: Increased survey efficiency is an important first step in animal monitoring. However, the inability to extract ADDs from the exponentially growing volumes of data is an ever-growing bottleneck for conservation agencies and a key limiting factor to implementing biodiversity-loss mitigation strategies. To overcome this problem, in 2017 we created a secure online research environment which uses machine learning to automatically find and identify animals/poachers in photo and video footage. Approved users from conservation agencies anywhere in the world can upload conservation images/videos to the website ( www.conservationai.co.uk) and almost immediately receive a robust list of the species in the footage. As of 21/9/20 we are supporting 61 users from 8 conservation projects across 4 different ecosystem types. The website has processed >35k images and detected over 100k objects.

Dr Theron, Head of Drylands Conservation Program at the Endangered Wildlife Trust: “… manually sorting images will take 10 to 15 times the amount of time required to simply upload the images to the machine learning pipeline.” [CS1]

Dr Piel, Director of the Greater Mahale Ecosystem Research and Conservation Project: “… drones coupled with machine learning can detect poachers 17x faster than when images are examined manually.” [CS2]

Impact 3 – Substantial savings in operating costs: Conservation agencies are typically charities operating to maximise ecosystem health on very tight budgets. Increased efficiency from Impact 1 & 2 translates directly into savings in operating and staff costs. Our conservation partners estimate the implementation of our technological solution saves them thousands of pounds per project per year.

Dr Hudson: our drone system represents “…a saving of £8,000, annually. In the field of conservation where funding for ‘routine’ actions such as monitoring is hard won, this saving is a fantastic achievement and will help to guarantee the sustainability of our field programmes, long-term.” [CS3]

Dr Theron: the increased detection efficiency, “… represents a significant amount of skilled resources that can be redirected to more productive conservation activities. The direct cost saving amounts to approximately £ 1500 per large camera trap survey (80 or more cameras) or more than 20 working days in one team members time.” [CS1]

Impact 4 – Transformative effect on biodiversity-loss mitigation strategies: The time and money saved can instead be invested in implementing optimal conservation strategies, which has a direct impact on halting biodiversity loss.

Dr Theron: “In summary, the collaboration with Prof. Longmore and Wich, has made our animal survey methods more cost-effective which allows us to focus more on other aspects of our work and will have a direct impact on our effectiveness.” [CS1]

Dr Hudson: “This new system has the potential to not only produce a step-change in the accuracy, frequency, robustness and geographic coverage of conservation data, but provide statistics of lemurs group numbers, and potentially even an individual’s size, weight, etc., revolutionising the potential for ecosystem management and tracking the efficacy of different restoration measures.” [CS3]

Impact 5 – Shaping global animal monitoring strategies: As a direct result of our research, leading conservation agencies around the world, such as the World Wildlife Fund for Nature (WWF), are reviewing their animal monitoring strategies to consider the systematic adoption of thermal-drone technology.

Nicola Loweth, Regional Manager, World Wildlife Fund for Nature: “In 2017, WWF approached Prof Longmore and Prof Wich … seeking assistance in resolving some of these challenge and stems from their research using (thermal) drones and machine learning to help conservation efforts. In particular we were interested to explore using these methods for our Bornean orangutan conservation work in Sabah (Malaysia) with an eye to potentially expanding these methods into other areas and species where WWF works around the globe.” [CS5]

Impact 6 – Shaping policy and legislation at a national level: With our collaborators in Madagascar we are advising the Madagascan Government’s Civil Aviation Authority as they implement a National Drone Policy document and build a national framework and regulations for operating drones. Dr Hudson: “… we expect this project to have long-term impacts at the highest national level.” [CS3]

Our research regularly appears in leading international media outlets (New York Times, BBC, Le Monde, National Geographic, etc.,) and has featured in 2 BBC Nature documentaries, reaching an estimated world-wide audience of 60 million in the last 2 years. This demonstrates the public reach and interest in the project, and has directly led researchers in conservation agencies around the world to contact us about collaborating to use our system.