From: mk_thisisit
The International Institute for Research on Artificial Intelligence (under the auspices of UNESCO) annually distinguishes global projects that significantly contribute to the implementation of the UN Sustainable Development Goals [07:10:06]. These goals, numbering 17, include targets such as the preservation of biodiversity and the protection of life on land [07:32:00]. Projects are selected for their positive impact on nature conservation and sustainable development [06:51:24].
The Empaza Application
A notable example of a project receiving this distinction is the Empaza application, developed by Poles at Apsilon in cooperation with scientists from the University of Stirling and national parks in Gabon [01:13:13]. Empaza was recognized among the top 10 applications globally by UNESCO, formally termed “outstanding” among the top 100 projects [07:52:00].
Purpose and Functionality
Empaza was created to help save endangered species, particularly in the forests of Central Africa [01:05:08]. The application utilizes modern technologies, such as machine vision and machine learning algorithms, to analyze images from camera traps placed deep within tropical forests [01:32:00]. These algorithms are trained to recognize various species present in the photos [01:41:00].
Camera traps, small devices the size of a palm, contain a camera, heat sensor, and motion sensor [02:22:00]. When an animal, like an African forest elephant, passes by, a photo is taken [02:30:00]. Ecologists and forest guards embark on expeditions, often lasting 3-4 weeks, to place and later collect these camera traps [02:07:00]. The collected data, typically around 30,000 photos per expedition, are saved on SD cards [03:02:00].
Traditionally, reviewing these photos manually took two to three weeks for a single expedition, as it is a tiring task requiring high attention to detail amidst dense forest imagery [03:35:00]. Empaza significantly reduces this time; it can process 30,000 photos in about 8 hours (one day) while running completely offline on a laptop [04:14:00]. The accuracy of the models is comparable to human assessment [08:59:00].
Impact on Conservation
The rapid analysis provided by Empaza allows scientists to quickly extract critical information, such as which camera trap photographed which animal and when [04:49:00]. This data can be mapped geographically to understand animal movements and assess changes in migration patterns, particularly those influenced by climate change [05:07:00].
Beyond migration studies, Empaza supports:
- Informing decisions on infrastructure development, helping avoid building roads in animal migration routes [05:48:00].
- Identifying gorillas in photos to assess their health status and track the spread of diseases within their communities, enabling faster intervention [05:59:00].
Accessibility and Funding
Empaza is an open-source application, with its full source code and machine learning models available on GitHub [06:27:00]. This open-source approach aims to maximize the positive impact on nature conservation by allowing as many scientists and national park guards as possible to use and develop the application [06:43:00].
The project was largely developed Pro Bono (pro-social), with some limited funding from non-governmental and governmental institutions over three years [09:06:00]. It is primarily used in national parks in Gabon, Cameroon, Kenya, and has recently been tested in Congo [06:56:00].
Broader Role of AI in Sustainable Development
Artificial intelligence plays a crucial role in sustainable development beyond business applications [12:04:00]. The UNESCO list of 100 projects highlights how machine learning can aid various sustainable development goals [12:27:00].
Other examples of AI’s application in environmental conservation include:
- A project in cooperation with the Arctowski station in Antarctica, where AI detects animals from drone images [13:01:00].
- Research at the North Pole helping to build models that accelerate the large-scale study of plankton, which is vital for ocean ecosystems and climate [13:21:00].
These initiatives demonstrate how AI can process vast amounts of data quickly, providing insights crucial for ecologists and other scientists working to understand and protect the planet [12:49:00].