[navigation]
Many people conflate face recognition with face detection and face tracking. While these technologies have a lot in common, they perform different functions and have different applications.
The term “face recognition” means matching the face in a photo or video to a specific person in the database. It is used for biometric identification, law enforcement, and the like.
“Face detection” is determining whether there is a human face in the photo or video and finding its precise location. And “tracking” means following it through each frame. These technologies are used in tandem for virtual try-on, video communication, social media, etc.
Banuba’s specialty is face detection and tracking, although the algorithms could be repurposed for recognition purposes. For example, one of our customers uses Banuba Face AR SDK as a foundation for their biometrics software now powering banking and cryptocurrency apps.
An SDK (software development kit) is a collection of libraries designed to be quickly installed in an application and perform specific tasks, e.g. recognizing and tracking faces. An API (application programming interface) is a set of rules and protocols that allows different software applications to communicate and interact with each other. If you need to access a third-party application, you can use its API to get the results quickly, even if you don’t know how the software works under the hood.
When making a facial recognition application, both SDKs and APIs serve a similar purpose – to help introduce a complex technology quickly and on a budget. Training neural networks to reliably recognize objects in different environments, lighting conditions, and video/photo quality takes time and money, as well as specialized knowledge. So many companies use ready-made tools instead of building everything from scratch.
These are the main advantages of using a facial recognition SDK or API:
Now, let’s cover the 10 most prominent facial recognition APIs on the market. Some of them already include everything, others are better used as a base for other facial recognition software.
Microsoft Azure AI Vision, formerly known as Microsoft computer vision API) is a powerful multipurpose computer vision service. It offers optical character recognition (OCR), image analysis, spatial analysis, and face recognition. This is what its Face API features:
The Face API access could be purchased separately, and includes 30.000 free transactions per month (calculated at the endpoint). Its pricing depends on the number of transactions, with larger packages receiving a bulk discount.
Face API’s common use cases include biometric identity verification, touchless access control, and face redaction. However, Microsoft places limitations on who can use this product: police departments in the USA and businesses who develop facial recognition technology for them are forbidden from using Face API, as a part of company’s ethical approach. Moreover, this product is only available to Microsoft managed customers.
Banuba is a premier artificial intelligence and computer vision lab from Europe. Its Face AR SDK uses state of the art AI to recognize human faces, but also has advanced features, e.g. TikTok-like masks and realistic virtual cosmetics. Several companies also use it as a foundation for biometric software that controls access to sensitive information and fights identity fraud.
This is what Face AR SDK has to offer:
Face AR SDK is compatible with native iOS and Android applications, as well as Flutter, React Native, Mac, Windows, Web, and Unity. It has a 14-day free trial that lets you test all the features. Moreover, its pricing is independent from the number of users, making it extremely scalable.
Rekognition is Amazon’s object detection service that includes a face recognition API. It is separated into two products, one for static images and one for videos. This is what they are capable of:
Rekognition has a complex pricing structure mostly based on the number of analyses conducted.
Cloud Vision API is Google’s image processing platform. It doesn’t have dedicated face recognition algorithms, but can serve as a reliable foundation for such software thanks to its efficiency. This is what it can offer in relation to faces:
Each part of Cloud Vision is priced separately depending on the number of uses. First 1000 units per month are free.
The numerous artificial intelligence and machine learning offerings of Clarifai include a face recognition API. It is commonly used for access control, authentication and verification, media exploitation, demographic analysis, and more.
Clarifai’s facial recognition technology can do the following:
Clarifai has a free version available for small developers. The paid plans also include 1000 free operations.
Inferdo has several image recognition solutions, including a facial recognition API. However, besides detecting and tracking faces, it can also locate facial features, determine age, and detect gender.
Inferdo has four pricing plans. Each of them except for the free one include a specified number of images to be processed. The excess ones are priced at $0.0005/img.
Kairos is one of the most significant facial recognition/face detection APIs on the market. It has gained popularity among both clients and investors, as the company secured over $11M in funding.
Its functionalities include:
Kairos has 4 cloud-based pricing plans, as well as an on-premise option. The subscriptions differ in the number of transactions per minute, as well as available features and a price per transaction.
Luxand is a premier face recognition API that counts several Fortune 500 companies among its customers. It includes both the core and advanced features:
Luxand has 4 pricing plans that differ in the number of transactions per second, faces in storage, and API requests.
Deepface is a Python library that claims to recognize faces better than humans do. It is free and open source, with the code available on GitHub.
Deepface can perform all the core face recognition and analytics functions:
Face++ is a Chinese product that has gained international recognition, customers, and awards. It is available as an API and as an SDK and includes:
API access can be bought on a monthly basis or using a pay-as-you-go system.
There is no single product that would be the best for all cases. Some business cases require a comprehensive boxed face recognition solution, others only need certain features or modules. The prices and pricing models also affect the selection process. The optimal option would be to sign up for a free trial and see the products for yourself.