Technology

The BinEye
Tech Stack.

Solar-powered edge compute. Cellular telemetry. AI inference on-device. A cloud data pipeline built for fleet scale. Every Bin Eye Bin ships with it included.

The Hardware

One module. Six jobs. Always on.

The BinEye Smart Module is a sealed, solar-powered edge compute unit factory-fitted to every Bin Eye Bin. It does its jobs simultaneously, without driver input and without a battery replacement programme.

Solar Power

Monocrystalline solar panel powers the module indefinitely. Integrated supercapacitor maintains operation through multi-day low-light periods. No battery programme. No scheduled replacements.

GPS Tracking

±2m accuracy. Cellular uplink. Location reports when needed rather than on a fixed clock. Full track history is kept for the life of the bin.

AI Camera Vision

2 MP wide-angle camera with IR cut filter. On-device inference using a lightweight CNN model updated via OTA, and it continuously learns. Classifies fill level, material type, contamination, and overfill from inside the bin.

Fill Monitoring

Time-of-flight distance sensor gives sub-centimetre fill depth, fused with the camera fill estimate. It learns each bin's activity, so a sheet thrown over the camera is not read as full. Configurable alert threshold — default 80%.

Cellular

Global SIM with automatic carrier selection. Works on local networks — automatic handoff keeps coverage across metro, regional, and remote corridor sites.

Filter Foot

The structural corner foot doubles as a filter housing. Liquid that collects in the bin drains by gravity through a replaceable cartridge that captures sediment and oil sheen, and a low-energy sensor reports what it sees. We are moving to a wired foot sensor, so there is no foot battery to replace.

OTA Firmware

All firmware updates delivered over-the-air via the cellular uplink. No site visits. No manual flashing. The AI classifier model is updated as the central model improves — every bin in the fleet gets smarter automatically.

The Data Pipeline

Edge to fleet dashboard in under 60 seconds.

Every telemetry event — fill reading, camera inference result, GPS fix, contamination flag — travels from the bin module through the cellular network to the BinEye cloud, is processed and classified, and surfaces in your fleet dashboard within one minute.

The pipeline runs on AWS ap-southeast-2 (Sydney). Data never leaves Australian jurisdiction without explicit customer consent. Operators can also access all telemetry via the REST API — for direct integration with BinManager, Xero, dispatch systems, or custom fleet software.

Build with Blues
01
Edge inference
Camera + ToF sensor → CNN model → event classification. All inference runs on-device before transmission. No raw images leave the bin by default.
02
Cellular uplink
Classified events + sensor readings + GPS position transmitted over 4G LTE. Encrypted in transit (TLS 1.3). Carrier-agnostic SIM auto-selects strongest available network.
03
Cloud ingestion
AWS ap-southeast-2 receives, validates, and time-stamps every event. De-duplication and gap-fill applied before storage. Full track history is kept for the life of the bin.
04
Fleet API
REST API exposes fill level, GPS, material classification, contamination flags, and collection history. Webhooks available for real-time push to fleet dispatch and billing systems.
05
Dashboard + alerts
Web and mobile fleet dashboard shows live bin status, route optimisation suggestions, and configurable alert thresholds. SMS and push notifications on overfill or contamination events.
06
OTA update loop
Model improvements and firmware updates deployed from cloud → device OTA. The entire fleet updates automatically — no field service visits required.

The AI Model

A classifier that learns from every bin in the fleet.

The BinEye waste classification model is a convolutional neural network trained on annotated bin imagery from Australian commercial and residential sites. It runs on-device, updating automatically as the central model improves.

Architecture

Lightweight CNN — on-device

Quantised convolutional neural network optimised for edge inference on the module's embedded processor. Sub-200ms inference time per frame. Runs continuously without impacting GPS or telemetry.

On-device · No cloud round-trip per inference
Classification outputs

5 material classes + fill level

General waste, recycling, green/organic, construction & demolition, and hazardous/contaminated. Fill level output is a continuous depth reading fused with the camera estimate.

5 classes · continuous fill depth · confidence score per event
Training data

AU-specific annotated imagery

Trained on annotated imagery from Australian commercial and residential skip bin deployments. Dataset grows with each deployed bin — every classification event (with operator confirmation) feeds back into the training loop.

Fleet-scale training loop · improves with every bin
OTA updates

Fleet-wide model deployment

Model updates are tested, validated, and deployed OTA to the entire fleet simultaneously. No field service. No manual update. Every bin gets the latest classifier automatically — improving accuracy fleet-wide each release cycle.

OTA · zero downtime · rollback capable

Security & Compliance

Australian data. Australian jurisdiction.

Data residency

All telemetry, imagery, and account data stored in AWS ap-southeast-2 (Sydney). No data stored outside Australia without explicit consent.

Encryption

TLS 1.3 for all data in transit. AES-256 for data at rest. Cellular SIM communications encrypted end-to-end.

Privacy Act compliance

Operated in accordance with the Australian Privacy Act 1988 (Cth) and the thirteen Australian Privacy Principles (APPs).

Access controls

Role-based access controls. Operators see only their fleet. Least-privilege IAM on all cloud infrastructure. Audit logging on all administrative actions.

No raw image egress

Camera inference runs on-device. By default, no raw images leave the bin. Event-triggered image capture for contamination and illegal dumping evidence can be enabled per customer.

Duty-free import

Delivered duty-free to Australia, New Zealand and Singapore. Enquiries from other ports welcome. Full documentation available at order confirmation.

The App

Your fleet, in your pocket.

Live bin status, location, fill and contamination alerts, and per-bin history — on iOS.

BinEye iOS app screen 1 BinEye iOS app screen 2 BinEye iOS app screen 3 BinEye iOS app screen 4 BinEye iOS app screen 5

// Pre-Orders Open — First Container Q4 2026

Every Bin Eye Bin ships
with this tech included.

Enquiries → View Bin Catalogue

BINEYE.AI PTY LTD · BETTERGROUP AI