University of Oxford Develops Facial Recognition Software for Wild Chimps
While facial recognition is not a particularly new technology, the advancement of artificial intelligence (AI) has revolutionized facial recognition. AI is embedded in almost everything nowadays from healthcare and social media to smart cars, banking, entertainment, and toys.
Our review of the Anki Cozmo Robot is one example of an AI-powered consumer device, aside from your average smartphone. When AI was embedded into facial recognition algorithms, an intelligent engine was created which gave rise to visual data software and visual content, which old facial recognition systems lacked.
This is what enabled University of Oxford researcher Arsha Nagrani and her fellows to develop an AI that tracks and recognizes chimpanzees in the wild.
The benefit of this technology is that it could make studying animals in the wild more efficient and accurate.
Using AI, researchers can cut down the time and resources needed to track animals in the wild and allow conservationists to study complex behaviors of chimpanzees and other primates.
The AI was trained using 50 hours of archival footage, spanning 14 years, from chimpanzees in Bossou, Guinea.
A total of 23 primates of all different ages were studied, which yielded 10 million facial images. The AI had an overall recognition accuracy rate of 92% and performed well under low-light and poor-quality images. Furthermore, when pitted against researches experienced in identifying chimps, the AI achieved 84% accuracy, taking 30 seconds to complete, while humans had an average accuracy of 42% and took 55 minutes.
By using the power of machine learning, it’s feasible to measure long-term behavior, like observing how social interaction in a group change over several generations.
This technology has the potential to be used for many things in the future, such as monitoring different species for conservation and help drive the adoption of AI systems to solve a range of problems in wildlife sciences. Researcher and DPhil student at Oxford’s Primate Models Lab Dan Schofield explained that the ability to closely monitor different species and populations using automated systems will be crucial for conservation efforts and behavioral research.
This not only highlights the many versatile uses of AI for primate research but how these advances can be applied in the human world.
With this in mind, the pace at which the development and application of AI are increasing in the US is moving rapidly, with law-enforcement, airport security, healthcare, and social media platforms, all using it to protect and enrich our lives. But MIT researcher Joy Boulamwini noted in her TED Talk how AI facial recognition algorithms are biased, and it’s the reason why the current technology is often less accurate for women and people with darker skin tones. This is a problem as implementing biased technology in high footfall areas like airports could generate a lot of false positives.
Boulamwini’s research, therefore, prompted AI scientists to improve their algorithms, especially as biometric facial recognition is becoming more widespread across a number of sectors.
With more students entering the field of AI research, it’s important that they keep in mind these biases and learn to avoid them. Online learning giant Udemy has more than 200 AI courses than help professionals ‘solve unique problems in the industry’ and shows just how disruptive AI technology is right now.
The learning platform’s range of courses are applicable to almost every major industry, including manufacturing, medical services, retail, and even music.
With the algorithmic bias issue at the forefront of research and the AI algorithms developed by the Oxford researchers, the potential for AI-driven facial recognition increases even further as professionals continue to develop their skills to solve these pressing issues.
Although it’s a little early to predict where facial recognition technology will be in the next few years, it will certainly become ubiquitous for law-enforcement in tracking criminals and other threats.
Aside from conservation, healthcare is the next big field with personalized humanoid, AI-driven robots that can help identify health symptoms, while taking on work from doctors so that they can focus on more serious cases.
The all-out market for AI-equipped cameras and software is predicted to be worth $7.76 billion in three years, and with the over 62 million security cameras in the U.S. alone, and millions more in everyone’s pockets, facial recognition will certainly become more widespread globally.