Computer Vision Applications for Wildlife Conservation

Species across the globe are experiencing climate change, habitat loss, and poaching, a critical combination of pressures that will lead to mass extinction without effective action. However, protecting threatened wildlife populations is a daunting challenge for governments and NGOs. Lack of resources and expertise mean that environments in need of conservation can often be inadequately monitored and protected.

Thankfully, AI developers are starting to introduce technology that can support conservation efforts. This important innovation is made possible by data annotation. Applying information to wildlife images and video helps machine learning models to grow and perform their given tasks with a high degree of precision.

This blog will look at three areas of wildlife conservation where computer vision is providing an important edge. In each case we will show how smart annotation provision, from companies like Keymakr, is supporting AI innovation.

Identifying animals

In order to protect wildlife researchers need to know as much information about animal populations as possible. This means counting population numbers as well as recognising individual animals. Traditionally this information is gathered using camera traps that take images when they are triggered by movement. Researchers trawl through thousands of these images to find target animals and gain a better understanding of population dynamics.

Of course this process is extremely time consuming, and can result in errors of identification. As a result conservation organisations are increasingly turning to AI developers to accelerate animal recognition. Google’s DeepMind team created a machine learning model that was able to detect and count a wide range of animals in the Sarengeti ecosystem, giving conservation efforts a vital boost.

Automated systems that are able to recognise and catalogue animals often rely on image annotation to power their algorithms. Professional annotation services can apply techniques such as point annotation to ensure that training datasets accurately reflect the specific details of individual animals, such as facial feature positions or specific markings.

Protecting habitats

Illegal poaching can decimate already declining wildlife populations. For example it is estimated that poachers kill 55 elephants every day, forcing this cornerstone species to the edge of extinction. Identifying poachers and catching them before they can act is extremely difficult due to the large, remote areas in which these animals roam.

AI applications can help to cover the gaps in animal protection by giving authorities a new surveillance tool. AI powered cameras can be set up strategically at points of access to wildlife areas. These cameras can monitor trails 24/7 and can alert anti-poaching staff to the presence of suspicious individuals in real time.

For these systems to work reliably they must be able to function in low light levels and in a variety of weather conditions. Image annotation services can help to assemble datasets which reflect the complex nature of natural environments. They can achieve this by leveraging data creation and collection expertise.

Monitoring ecosystems

Preventing habitat loss and securing fragile ecosystems is at the core of wildlife conservation. In order to assess the health of crucial forest habitats researchers have turned to AI. Machine learning can allow models to interpret terabytes of satellite imagery and create forest inventories for entire continents.

These images can contain information on dominant tree species or the total amount of carbon contained in forested areas. This kind of large scale data can allow conservationists to target areas of habitat loss, and help to guide authorities’ decision making around logging and land use.

Smart data annotation can help to widen access to technology of this kind. Satellite image annotation, carried out by skilled operators, can support innovators in this field by providing access to precisely labeled training data at the large scales that models demand.

Keymakr is a professional annotation service with expertise in data collection and annotation.