From: mk_thisisit
The emaza application, developed by Poles, is a groundbreaking tool designed to aid in the preservation of endangered species, primarily within the vast forests of Central Africa [00:00:00], [00:00:49], [00:01:08]. This innovation was recognized globally, being listed among the top 10 applications worldwide by UNESCO [00:00:40], [00:00:52], [00:07:06].
The Challenge in Rainforests
Ecologists and forest guards undertake expeditions into dense forests, some as vast as Poland, to monitor wildlife [00:00:08], [00:02:12]. A key method involves deploying camera traps – small devices equipped with cameras, heat sensors, and motion sensors – that automatically capture images when an animal passes by [00:02:20], [00:02:23], [00:02:25], [00:02:27]. After several weeks, scientists collect the SD cards from these traps, often accumulating around 30,000 photos per expedition [00:03:02], [00:03:19].
Manually reviewing such a massive volume of images is incredibly time-consuming and arduous. It typically takes two to three weeks for humans to classify 30,000 photos [00:03:35], [00:03:37]. The task is tiring, leading to a significant drop in efficiency and accuracy over time, especially given the complex forest backgrounds filled with bushes, branches, and sun spots [00:03:43], [00:03:46], [00:03:49].
Empaza’s Solution: Leveraging Machine Learning
The emaza application leverages machine learning algorithms, specifically machine vision or computer vision, to automate the identification of species in these photos [00:01:34], [00:01:36], [00:01:39], [00:01:43]. Crucially, the application operates completely offline, allowing scientists to process data directly on their laptops in remote bases without internet access [00:03:24], [00:04:17].
With emaza, a batch of 30,000 photos can be processed in approximately 8 hours, or about one day, a dramatic reduction from the 15 days (three weeks) required for manual review [00:04:39], [00:04:42], [00:04:44]. While no machine learning model is perfect, emaza’s accuracy is comparable to human classification, and the errors balance out, leading to robust ecological models [00:09:01], [00:09:11], [00:09:13], [00:09:17], [00:09:20].
Impact and Benefits
The rapid processing allows ecologists to quickly extract vital information and move to the next stage of analysis [00:04:49], [00:04:51]. This includes:
- Biodiversity Preservation: By identifying which animals appeared where and when, scientists can map their movements and assess species populations [00:05:01], [00:05:04], [00:05:44], [00:05:46]. This data informs decisions like avoiding road construction in critical animal migration routes [00:05:51], [00:05:54].
- Monitoring Animal Migration: The application helps track changes in animal migration patterns, such as those of African forest elephants, which are being drastically altered by climate change [00:05:07], [00:05:10], [00:05:12], [00:05:14], [00:05:16]. This information would otherwise be unobtainable [00:05:27], [00:05:28].
- Disease Tracking: Empaza has been instrumental in studying diseases, such as one spreading in gorilla communities. It quickly identifies photos containing gorillas, allowing experts to assess their health status and understand disease spread, enabling timely intervention [00:06:01], [00:06:02], [00:06:04], [00:06:06], [00:06:11], [00:06:15].
Availability and Recognition
Empaza’s full source code and machine learning models are open source and available on GitHub [00:06:27], [00:06:29], [00:06:32], [00:06:35]. This open approach aims to maximize its positive impact on nature conservation by allowing as many scientists and park guards as possible to utilize it [00:06:43], [00:06:45], [00:06:47], [00:06:49], [00:06:51].
The application is primarily used in Gabon and has seen use in Cameroon, Kenya, and recent testing in Congo [00:00:22], [00:00:25], [00:00:26], [00:00:28], [00:06:57], [00:07:01], [00:07:04].
Empaza received significant recognition by being distinguished as one of the “outstanding” projects (effectively a Top 10) among the 100 global projects highlighted by the International Institute for Research on Artificial Intelligence (under UNESCO auspices) [00:07:10], [00:07:11], [00:07:14], [00:07:17]. These projects contribute most to the UN Sustainable Development Goals, specifically “preservation of biodiversity” and “protection of life on land,” which are directly supported by emaza [00:07:27], [00:07:29], [00:07:32], [00:07:34], [00:07:36], [00:07:39], [00:07:41].
Origin and Development
The idea for emaza emerged in the early 2010s when an ecologist from the University of Stirling in Scotland, researching tropical forests in Gabon, sought to automate camera trap data processing [00:08:08], [00:08:11], [00:08:15], [00:08:17], [00:08:20], [00:08:22]. The Apsilon company, specializing in machine learning, developed the models and application after realizing their expertise could best assist the ecologist [00:08:41], [00:08:47], [00:08:50].
The project was largely pro bono, created as part of an internal “data for gold” program [00:10:13], [00:10:15], [00:10:16], [00:11:41], [00:11:44], [00:12:00], [00:12:02]. While some small fragments of development were financed by non-governmental and governmental institutions over three years, it remains primarily a pro-social effort [00:11:50], [00:11:52], [00:11:54], [00:11:56], [00:11:58] oftentimes through the company’s computer vision and image analysis work that brings in commercial income [00:11:31], [00:11:33], [00:11:34], [00:11:37], [00:11:40].
Broader Role of Artificial Intelligence in Conservation
AI plays a crucial role in global sustainable development efforts beyond just rainforest protection [00:12:06], [00:12:09]. Many projects utilize machine learning to assist with sustainable development goals [00:12:34], [00:12:36], [00:12:38]. Scientists, particularly ecologists, require rapid analysis of large datasets [00:12:43], [00:12:45], [00:12:47], [00:12:49].
For instance, the Apsilon company also collaborates with the Arctowski station in Antarctica, a Polish Academy of Sciences station, to detect animals from drone images [00:00:30], [00:00:32], [00:00:35], [00:13:01], [00:13:04], [00:13:06]. Similarly, at the North Pole, they assist in building models to accelerate the study of plankton on a large scale, which is crucial for oceanic life and climate impact [00:13:21], [00:13:22], [00:13:25], [00:13:27], [00:13:30], [00:13:32].
While Poland has had other applications recognized by UNESCO in previous years, emaza stood out in its category [00:13:36], [00:13:39], [00:13:42], [00:13:44].