VISION AI SOLUTION

NEXVDOVIDEOAI

Vision AI from camera to decision — capture, on-device analysis, and a documented path that takes your model from weights to a product you can ship.

SCROLL

THREE WAYS IN

Each path runs the same SDK surface. Pick the hardware that fits your build.

JETSON

SmartVDO Edge systems

Jetson systems from Jetson Thor to Orin Nano — cameras in, decisions out at the edge.

Full lineup at yuan.com.tw

Talk to an engineer →
INTEL X86

Capture cards + x86

PCIe capture density for servers and industrial PCs running the same SDK surface.

Compatibility matrix — pending publication

Talk to an engineer →
DEV KIT

Pandora GenAI dev kit

The bench kit for proving your pipeline before you commit a platform.

Kit contents — pending publication

Talk to an engineer →

CONTINUOUS DUTY

Built for the hours nobody watches

Recording is the part of the pipeline that cannot drop a frame at 3 a.m. NexVDO's media plane comes from three decades of capture engineering, where long-duration recording is the baseline, not the stress test.

The same discipline carries into analysis: pipelines designed to run at the line, on the pole, and in the rack — for as long as the deployment does.

Illustrative footage — generated media, see claim register.

From SDK to running system

Four surfaces you touch between unboxing and production.

Capture

Enumerate devices, bring sources online, and see frames on screen in the first session.

Deploy models

Bring your own weights through the documented QDEEP path and run them against live video.

Manage the fleet

Device and fleet management is answered through the Allxon partnership.

Review & export

Scrub recordings, pull clips, and move evidence out of the box on production paths.

See what the pipeline sees

CAMERA FEED
ANALYSIS VIEW

Overlay visualization is illustrative — generated media, not SDK output. SDK output samples ship with evaluation access.

4 modulesone SDK surface
2 platformsNVIDIA Jetson & Intel x86
48 hengineer answer on evaluations

QDEEP

Your weights, deployed

QDEEP is the documented custom-weights path in the Analysis module: bring the model your team already trained, and run it against live video on-device. No forced retraining, no black-box conversion service — a path an engineer can read, follow, and argue with.

INPUT

Your trained weights, in the formats your team already uses.

PATH

A documented conversion and deployment flow — published with evaluation access.

OUTPUT

Live inference in the same pipeline that captures, records, and streams.

Why teams pick NexVDO

01

One SDK surface

Capture, Record, Stream, Analysis — one API across Jetson and x86.

02

Your model, your weights

QDEEP is a documented path, not a conversion black box.

03

30 years of capture

The media plane comes from three decades of video engineering.

04

Inside the NVIDIA ecosystem

NPN Preferred Partner 2025; the Edge system line is built on Jetson.

05

Engineers answer

Evaluation requests get a substantive answer within 48 hours.

06

Included with hardware

The SDK ships free with YUAN hardware — no license line.

Need a human?

Engineers answer evaluation requests within 48 hours. Business and hardware questions route to Neteon — YUAN's U.S. front door.

Evaluate on real hardware

Request an evaluation and get a substantive answer — from an engineer — within 48 hours. Evaluation hardware ships from U.S. stock.

Request an evaluation

Evaluation form ships with the full site build — request path pending.