Nexariza AIOverview
Detectra AI unifies object detection, multi-object tracking, speech transcription and on-screen text recognition into a single pipeline. A cross-modal transformer with 8-head attention aligns visual, audio and text signals to power seven specialized AI services.
Understanding video means fusing several modalities — objects, motion, speech and on-screen text — that most systems handle in isolation.
A unified multimodal pipeline (YOLOv8 + ByteTrack for vision, Whisper for speech, EasyOCR for text) aligned by an 8-head cross-modal attention transformer, exposed as seven specialized services.
Technology
YOLOv8 · ByteTrack · Whisper · EasyOCR · CrossModal Transformer · Python