I found this :

https://www.fujitsu.com/global/about/resources/news/press-releases/2019/1015-01.html

Looks like it was posted about a month ago, I haven’t really found much more info on it, was trying to see if it was anything groundbreaking, the details provided are slim to none. Just gives a basic overview of how neural networks seem to work, and some fancy terms for normalization/transformations/landmark detection. Thanks.

They make some bold claims:

This technology has achieved a high detection accuracy rate of 81% even with limited training data.

This technology is also more accurate than other existing technologies according to certain facial expression recognition technology benchmarks