How Image Annotation Enables Facial Recognition in a World of Masks

Masks are now mandated or recommended in most countries of the world. In an effort to reduce the spread of Covid-19, health authorities across the globe have made face coverings a cornerstone of their pandemic response. These interventions are well supported by science and should be welcomed as a low impact way of moderating the worst impacts of the virus.

However, the widespread use of facial coverings has produced a number of second order effects, one of which is in the field of facial recognition AI.

By obscuring facial features, masks have the potential to slow the development of facial recognition technology and hamstring already existing systems. In order to overcome this emerging issue, AI companies must work with experienced image and video annotation providers, like Keymakr, to ensure that the training data they receive is representative of a changing world. This blog will address some of the facial recognition use cases affected by face masks and then discuss how these problem areas are being supported by data annotation.

The Importance of Facial Recognition AI

Facial recognition technology is becoming increasingly important in a number of industries, allowing businesses and organisations to gain access to the information they need faster and more efficiently:

• Security: There are many applications for facial recognition AI in the security realm. This technology is being employed to identify potentially dangerous individuals at airports and military installations. Smart security cameras are also able to identify emotions and alert security staff to potential risks in public spaces.

• Retail: In the retail sector facial recognition is allowing businesses to analyze facial expressions to understand customer sentiment towards the promotions and products on shelves. AI powered monitoring can also reward frequent customers with loyalty based deals and provide them with personalised product recommendations. It is also enabling the development of smart checkouts that can aid with loss prevention strategies and identifying potential shoplifters.

• Autonomous vehicles: Facial recognition technology is now being employed for in-cabin AI. This enables systems within a vehicle to recognise when a driver is falling asleep, or exhibiting any other potentially dangerous behaviour at the wheel. These computer vision based models have the potential to improve driver training and make the roads safer for all of us.

All of these potential and current applications of facial recognition capable AI are complicated by the widespread and continuing use of masks. Computer vision AI is reliant on the training imagery and videos that teach it how to interpret the world around it.

If images used to train a facial recognition AI do not feature masks then it is unlikely that the model will be able to function when presented with masked individuals in the real world.

Overcoming the Challenge of Masks Through Collaboration

This is where professional data annotation services come in. Keymakr deploys teams of experienced and attentively managed annotators working together in one location. This facilitates the creation of high quality labeled datasets, and allows for responsive and accurate troubleshooting and collaboration.

In the case of a developing situation like mask mandates, a service like Keymakr is able to rapidly develop bespoke image and video annotations featuring masks in a plethora of contexts.

By labeling and tracking facial features above the nose professional annotation services are establishing datasets that will support new iterations of facial recognition AI, that are capable of functioning in a world shaped by Covid-19.

Keymakr is your image annotation solution. Take advantage of cutting-edge tools and an experienced and professionally managed team of annotators. Contact a team member to book your personalized demo today.