Technology

Systems built for
real access.

Every product in development at SignaVision targets a specific gap between existing technology and the needs of Deaf communities. No padding. No fluff.

Computer Vision · LTI based Sign Language Recognition

Sign Language
Recognition
via LTI 1.3

AI models built on computer vision that interpret visual signing in real time. The system analyzes video input frame-by-frame, extracting gesture, handshape, movement, and spatial position to construct meaning.

Unlike speech-to-text systems, sign language recognition requires understanding three-dimensional space through a two-dimensional lens. SignaVision develops models trained for this specific challenge.

The primary deployment of this system is inside LMS platforms, delivered via LTI integration. A student signs a response. The system observes it in real time, evaluates it against a defined answer, and reports the result back to the gradebook — without video leaving the session, without credentials crossing system boundaries.

LTI 1.3 — focused on interaction and communication within the LMS, not just grade passback.

Frame-accurate gesture extraction
Real-time inference on dedicated GPU hardware
WebRTC video stream compatible
LTI 1.1 / 1.3 integration — Canvas, Blackboard, Moodle, D2L
Student responses evaluated and scored within the LMS session
Ongoing model improvement pipeline

Recognition → Assessment Pipeline

Student Signs Response (WebRTC / Camera)
Frame Extraction & Preprocessing
Pose Estimation Model
Gesture Classifier — landmark input, letter output
Boundary: Session-bounded assessment engine
LTI grade passback — explicit outcome URL only
LMS gradebook updated — session sealed

Live Caption Stream

Incoming audio signal
Speaker: "The session will begin shortly—"
— please ensure your camera is enabled.
Transmitting captions in real time…
Real-Time Captioning

CaptionOnFly

Real-time caption generation designed for presentations, classroom sessions, and learning environments. CaptionOnFly processes live audio or media and delivers synchronized captions directly to participant screens.

Built for accessibility, not as a plugin — CaptionOnFly integrates at the platform level, giving educators and learners captions without the friction of third-party tools or manual CART requests.

Sub-second caption latency
LMS-embeddable via LTI
Speaker identification support
Video-First Communication

Accessible Forums

Discussion platforms redesigned for sign language users. Instead of typing text responses, participants record short video replies — sign language becomes the natural medium, not an afterthought.

Traditional forums were designed around text. Accessible Forums are designed around visual communication from the ground up — the same principle that should have been there from the start.

Video responses as first-class content
AI-generated captions on video submissions
LTI integration with LMS gradebooks

Forum Thread

Instructor · Video Response
Instructor video response

ASL · 0:34 · Auto-captioned

Student · Video Response
Student video response

ASL · 0:18 · Auto-captioned

Interoperability

LMS integration without compromise.

SignaVision tools connect directly to existing LMS ecosystems using LTI standards. No shadow systems. No duplicate accounts. Tools appear inside the platforms educators already use.

Canvas

Blackboard

Moodle

D2L / Brightspace

Want to see these in action?

Contact us about pilot programs, research partnerships, or integration discussions.

Get in touch Visit the Lab