SEWA (2015–2018) targeted automatic sentiment estimation from spontaneous, unconstrained video — the direct technical foundation of Realeyes' commercial platform.
REALEYES KFT
Hungarian emotion AI SME measuring facial expressions and sentiment from video; specialist contributor to biometrics and affective computing research.
Their core work
Realeyes is a Budapest-based technology company specialising in emotion AI — the automated measurement of human emotional and attentional responses through computer vision analysis of facial expressions. Their core capability is real-time detection of sentiment, engagement, and affect from video input, applied to both behavioural research and identity verification. In EU research consortia they contributed proprietary facial analysis technology: to SEWA for sentiment recognition from spontaneous, in-the-wild video, and to SpeechXRays for multimodal biometric verification combining face, lip movement, and speech signals. Commercially, Realeyes operates an emotion measurement platform used in advertising and media research to gauge audience reactions at scale.
What they specialise in
Both SEWA and SpeechXRays required automated analysis of facial action units and expressions, reflecting the central capability Realeyes brings to any consortium.
SpeechXRays (2015–2019) fused acoustic speech analysis with machine vision of lip movement and facial expressions into a single biometric authentication pipeline.
Both projects operate at the intersection of spontaneous human behavioural signals and machine learning inference, the applied research substrate behind Realeyes' products.
How they've shifted over time
Both H2020 projects began in 2015, which limits a meaningful early-versus-late comparison within this dataset alone. What the two projects reveal together is a natural extension from pure affective computing — SEWA's sentiment estimation in everyday, unconstrained settings — toward security-relevant identity verification in SpeechXRays. This suggests Realeyes deliberately repositioned its facial analysis technology beyond market research into regulated identity and access control domains. Their H2020 participation ends in 2019, and whether they continued EU-funded research after that point is not captured here.
Their project arc points from consumer-facing emotion measurement toward biometric security applications — organisations building identity verification or behavioural authentication systems would find a natural technology fit.
How they like to work
Realeyes participated exclusively as a third party in both projects — never as formal coordinator or named partner — indicating they contribute a specific, deployable technology component rather than shaping research direction. This pattern is typical of commercial SMEs that integrate a production-grade platform into consortia led by universities or research institutes. With 16 unique partners across just 2 projects, they operated in moderately large, multi-national teams consistent with ICT Innovation Actions.
Realeyes has connected with 16 unique consortium partners across 6 countries through only 2 projects, suggesting exposure to a broad but not deep European research network concentrated in ICT and security. Their partners are likely a mix of universities and technology companies active in computer vision, speech processing, and biometrics.
What sets them apart
As a commercial SME, Realeyes brings production-grade emotion recognition software to research consortia — a rare combination of academic-level measurement accuracy and a deployable product with real market validation. Most academic partners in affective computing work with research prototypes; Realeyes' technology is commercially tested in advertising and media markets, offering a shorter path from research outcome to live deployment. For a consortium that needs real affect or identity signals as an input layer, they reduce integration risk considerably.
Highlights from their portfolio
- SpeechXRaysThis Innovation Action extended Realeyes' emotion AI into biometric identity verification — a regulated security domain — by fusing speech acoustics, lip movement analysis, and facial expressions into a single authentication pipeline, showing the cross-sector reach of their core technology.
- SEWATargeting sentiment estimation from spontaneous, uncontrolled in-the-wild video directly validated Realeyes' commercial capability in a rigorous EU research context, lending academic credibility to their market-facing platform.