Εμορεκο : ενίσχυση της κατανόησης χρήστη μέσω συναισθηματικής ανάλυσης πολυμέσων
Emoreco : enhancing user understanding through emotional analysis of multimedia content

View/ Open
Keywords
Emotion recognition ; JavaFX ; Desktop analytics ; SQL Server ; Google Cloud Vision ; Video analysis ; ETL ; HCI ; Medallion architecture ; Star schema ; Dimensional modelingAbstract
Emoreco is a native JavaFX-based desktop application for end-to-end emotion recognition workflows in research and applied analytics. This solution integrates local data management with cloud-based perception, facilitating consistent experimentation, prototyping, and exploratory analysis. Emoreco connects to Microsoft SQL Server for structured storage and can call external services, including an Emoreco web API and Google Cloud Vision, enabling hybrid on-device/remote inference. The architecture emphasizes transparency and reproducibility through file-based configuration, explicit dependency management (Maven), and a separable model-view-controller structure.
Data architecture is a first-class concern in Emoreco. Ingested media, telemetry, and logs land in a Medallion Architecture: Bronze tables retain raw capture artifacts and service responses; Silver tables standardize and denormalize detections with consistent timestamps, identities, and quality flags; Gold tables provide analytics-ready aggregates and curated views for reporting and export. On top of this lakehouse-style layering, the analytics subsystem follows a Star Schema: a central Results fact (frame-/segment-level inferences with confidence scores) is joined to conformed dimensions for Video (source, resolution, fps), User/Session (operator, cohort), Time (hierarchies to second/frame), Emotion Label (taxonomy and mappings), and Parameters (model, thresholds, preprocessing). This combination preserves lineage and auditability while enabling performant queries even when dataset size increases exponentially.
Keywords: emotion recognition, JavaFX, desktop analytics, SQL Server, Google Cloud Vision, video analysis, human-in-the-loop, reproducibility, ETL, HCI, Medallion Architecture, Star Schema, dimensional modeling


