Net-Takeaway Engine · Vision Model for Unmanned Retail

Vision, settled.
True north for unmanned retail.

Polaris is a vision foundation model built for smart vending cabinets. From door-open to door-close it tracks every take and every put-back, then outputs a net-takeaway list you can invoice — not what the camera saw, but what the customer owes.

Order Session Cabinet CAM 01
Take0
Putback0
Net0
  • 01 ×Sparkling water 500 ml¥ 6.00
  • 01 ×Salted-egg potato chips¥ 8.50
Settlement ¥ 14.50
Audit · Clear Settled in 1.6 s
Illustrative output
Input Order-session video Delivery API / Edge-cloud Settles on Net-takeaway list Fallback Cloud audit loop
01 / Live Demonstration

One real order video.
Every judgment the engine made, written on it.

Raw in-cabinet footage with Polaris's visualization overlaid: tracking boxes, event states and the running take / putback / net counters. Nothing staged — this is the engine's original output.

Cabinet CAM · Order Session Polaris Engine · Overlay On
Hand–Object Tracking Event State Machine Take / Putback / Net
1280 × 720 · 25 fps · Unedited engine output Final: take = 1 · net = 1

Three layers of information on screen

Every overlay element corresponds to one core mechanism of the engine.

A

Tracking boxes · Hand–object association

A product only moves because a hand moves it. The engine tracks hands and products jointly; only hand-driven motion can trigger an event, so rack vibration and lighting shifts never become transactions.

B

State-machine text · Occlusion keep-alive

When a product is hidden behind a hand or body, its box is kept alive and the event state machine keeps reasoning — until the item is visible again or has clearly left the cabinet. Out of sight is not out of the order.

C

Counter bar · Net-takeaway settlement

Takes and put-backs accumulate in real time; net = take − putback. The count is final the moment the door closes — that one line of numbers is what the settlement is based on.

D

Count first, then recognize

Once the net count is fixed, SKU recognition only has to answer "which products are these" — on the most confident frames. The output is a settlement list ready for invoicing; uncertain orders route to the cloud audit loop.

02 / Core Capabilities

Four mechanisms. One auditable settlement chain.

Net-Takeaway

Net-takeaway recognition

Hand–object tracking plus an event state machine accumulate takes and put-backs event by event. The net result is fixed at door-close. What gets settled is behavior, not pixels.

SKU Recognition

Post-hoc SKU recognition

First fix how many items left the cabinet, then identify each one. Recognition effort collapses onto a few key frames and a known count — stable settlement output even against a 10,000-SKU library.

Occlusion Robust

Occlusion & anomaly robustness

Boxes stay alive through occlusion and identities are never dropped. The physical prior — products move only with hands — filters out rack collapses, lighting jumps and every other non-transaction disturbance.

Audit Loop

Cloud audit loop

Low-confidence orders route automatically to an audit channel; cloud review writes the settlement back, and reviewed samples flow into training. Every hard order makes the model sharper.

0%+
Net-takeaway count accuracy
Net-Takeaway Accuracy
0+
SKUs in production library
SKU Library
<0s
Settlement latency per order
Settlement Latency
0%
Reduction in manual audit cost
Audit Cost Reduction

* Illustrative targets — ask us for the measured evaluation report.

03 / Pipeline

From one video to one invoiceable transaction.

Six stages, fully traceable. Every intermediate conclusion of every order can be reviewed.

  1. 01 / INPUT

    Order video

    The in-cabinet camera records door-open to door-close. One order, one clip.

  2. 02 / TRACK

    Hand–object tracking

    Hands and products are associated frame by frame: who moved what.

  3. 03 / FSM

    Event state machine

    Takes, put-backs and occlusion keep-alive resolve each item's final fate.

  4. 04 / NET

    Net-takeaway list

    net = take − putback. The count is final when the door closes.

  5. 05 / SKU

    SKU recognition

    Post-hoc recognition confirms each item's identity and price.

  6. 06 / API

    Settlement API

    The settlement list ships via API; hard orders go to cloud audit review.

Swipe to see all six stages →

04 / Solutions

One engine. Four ways to put it to work.

For Cabinet Manufacturers

Ship every cabinet with settlement built in.

An edge-cloud package engineered into the cabinet itself: camera selection, mounting positions and calibration are all part of the delivery, with a clean split between on-device inference and the cloud engine.

  • Vision replaces gravity sensors and RFID — a leaner bill of materials
  • Full integration support shortens the productization cycle
  • Recognition upgrades ship OTA, so cabinets keep appreciating after they ship
For Retail Platforms

One engine behind every cabinet on your platform.

The recognition API plugs into your platform's settlement system and serves mixed cabinet fleets with one consistent standard of accuracy and evidence.

  • Uniform recognition quality across cabinet makes and models
  • Per-order billing that scales with platform volume
  • Video evidence chain standardizes dispute handling platform-wide
For Large Operators

Turn shrinkage disputes into evidence-backed line items.

Every order carries a net-takeaway list and a video evidence chain. Disputed orders trace back in minutes, and the loss column in your P&L finally itemizes.

  • Works on existing cabinets — no hardware retrofit required
  • Hard orders auto-route to audit; review teams get dramatically leaner
  • Per-order pricing amortizes with scale
For Independent Operators

Enterprise-grade recognition, pay as you go.

If you run your own system, the settlement API drops straight into it — integration measured in weeks, not quarters, with the cloud audit loop included from day one.

  • REST API with clear docs — integrate with a small team
  • Pay per order, no minimum commitment
  • The same audit loop and evidence chain the big fleets get
Get Started

Bring one real order video.
We'll settle it in front of you.

A 30-minute technical demo on footage from your own cabinets — accuracy and latency measured live, end to end.

polaris@beessell.com · 4008-321008