The development is based on lightweight neural network models, including MT-EmotiMobileFaceNet and MT-EmotiEffNet. The algorithm uses a method to smooth predictions using averaging and Gaussian filters, which improves recognition accuracy by 4.5 percentage points based on the F1 measure.
The system is capable of simultaneously detecting facial expressions, emotional valence and intensity, and analyzing facial muscle activity using the Paul Ekman system. The technology ensures information privacy by working directly on mobile devices without transferring data to the cloud.
The development placed second in the international Composite Expression Recognition competition at the ECCV 2024 conference. The source code of the models is available in the open library EmotiEffLib for further development of the technology.
Potential applications of the system include digital marketing, monitoring driver condition, and creating diagnostic tools in psychology.
Source: Ferra
I am a professional journalist and content creator with extensive experience writing for news websites. I currently work as an author at Gadget Onus, where I specialize in covering hot news topics. My written pieces have been published on some of the biggest media outlets around the world, including The Guardian and BBC News.
