Exploring the Development of "Face Recognition" Technology in Nodeflux TechTalk

Last week (20/11), artificial intelligence-based service developer startup nodeflux held an interactive discussion forum entitled "Nodeflux TechTalk". The goal is to share experiences and knowledge about trends in using AI for various needs. This inaugural event took the theme "How We Engineered Our Face Recognition and Bring It to World-Class Standard".

There were two speakers presented, namely AI Research Group Lead Nodeflux Alvin Prayuda and Engineering Manager Nodeflux Ilham Imaduddin.

Through his presentation, Alvin explained how the deep learning algorithm (deep learning) used in the development of facial recognition platforms (facial recognition). Technically, applying this method produces a unique set of numbers that represent the characteristics of the scanned face object. Other methods used in face recognition is embedding, providing efficiency for mass face recognition.

"Deep learning Dalam face recognition there it is pipelinenya, where we initially capture the face image, then determine landmarkand generate a unique number that represents this person, which in the end defines the face as the face of the X," explained Alvin.

Followed by the Ilham session which discussed details about the method embedding. He started by explaining the challenges of its implementation, namely in accelerating the process and managing large data flows. This can be anticipated with a computer system that can work efficiently, supported by an effective representation of the algorithm in code.

He also shared the benefits, "From previously it could take almost 43 seconds to process data, with this new method, this process has been accelerated many times. And one of our achievements is being able to identify faces from 150 million data. facial embedding in 60 milliseconds. We're working with NVIDIA to look for other ways to optimize AI for better benefits."

Forms of technology implementation face recognition can reach various areas, including being used for public security and safety solutions. For example, it is applied real-time to process data from CCTV, attendance system, and smart building.

Seeing the high enthusiasm, in the future Nodeflux TechTalk events will be held regularly and continuously, and open to all stakeholders from the government, universities, private sector, community, and media to share experiences and knowledge about AI.

--

Disclosure: DailySocial is a media partner of Nodeflux TechTalk