Skip to main content
CONPDS TRACKER

On-Premise Container OCR Infrastructure

The evidence source for every container that passes a camera — on-premise, GPU-powered, reading container numbers from any IP camera, and publishing container-linked reads in open JSON via MQTT. The OCR engine that powers ConPDS Autogate, and a standalone evidence source for any RTSP camera writing into the same container-linked record chain Checker writes to. Your hardware. Your data. Your integration.

Operating at container depots, terminals, and logistics facilities across Europe.
Read real customer outcomes →
No obligations — reply within 1 business day

What Is Container OCR Infrastructure?

On-premise container OCR infrastructure is the evidence source underneath the operation: it connects to IP camera streams, uses computer vision to detect and read ISO 6346 container numbers as containers pass, and publishes a container-linked record (number, camera, lane, timestamp, confidence) as structured JSON. Downstream systems — gate logic, the depot management system, the operational evidence layer — act on the same record without manual data entry. The same OCR engine powers ConPDS Autogate at the gate AND stands alone as an evidence source for any RTSP camera in the yard, the workshop, or the rail/Ro-Ro lane.
🏗️ Container Depots 🚂 Intermodal Terminals 🚢 Ro-Ro Terminals 🏭 Warehouses ⚓ Port Facilities
ConPDS Tracker
videocam 📷 Capture Multiple RTSP camera streams, grouped per lane, zone, or checkpoint
memory 🧠 Process Motion-triggered deep-learning OCR powered by ConPDS Core engine
send 📡 Publish Open JSON via MQTT — per camera or per camera group
Customer Integration
notifications_active 📥 Subscribe Custom code subscribes to the MQTT broker and receives OCR data from Tracker
sync_alt ⚙️ Process Data processed according to your custom algorithm and business rules
task_alt ✅ Act Update your systems — gate logic, DMS, TOS, WMS, API, database, or any platform
verified
ISO 6346 / BIC / ISO TankHorizontal & vertical, single- & double-line
bolt
As low as 30msGPU-accelerated processing per frame
lan
No DMS dependencyWorks with any management system
dns
On-premiseYour hardware, your network, your data

Where Manual Container Number Capture Quietly Costs Money

person
60 moves / day
Each movement manually read, typed, and verified by a gate officer — throughput ceiling tied to staff availability
error_outline
1 typo
A single transposition error creates a data mismatch that may not surface until a dispute is raised weeks later
gavel
Weeks later
Documentation mismatch discovered during a claim — no verifiable record to resolve the dispute

Without Automated Container Monitoring

Container numbers entered manually — transcription errors create downstream data mismatches
Gate officer required for every movement — throughput ceiling tied to staff availability
No timestamped record of which truck carried which container at which moment
Data lives in paper logs or spreadsheets — not available to downstream systems in real time
Integration with DMS requires manual re-entry — doubling the error risk

With ConPDS Tracker

Container numbers read automatically from camera streams — no manual entry
Up to 99% OCR accuracy — results published as JSON within 30ms of capture
Every read timestamped and attributed to camera and lane — full movement audit trail
Results available to any MQTT subscriber immediately — DMS, custom logic, database, alerts
Gate officer freed from data entry — focus shifts to exception handling

✗ Before: Manual Container Number Entry

A damage dispute is raised three months after gate-out. The gate register shows a container number — but a transposition error means it does not match the container that actually departed. There is no camera evidence, no timestamp, and no way to reconstruct what happened. A disputed charge without defensible evidence becomes a direct operational cost.

✓ After: ConPDS Tracker

The same dispute is raised. Tracker's OCR log shows the exact container number read from the camera at the moment of movement, the camera and lane, and the timestamp. The record is human-untouched, container-linked, and produced on demand — defensible by default.

How ConPDS Tracker Container OCR Infrastructure Works

01
Camera Connection
Tracker connects to IP cameras via RTSP
02
Motion Detection
Vehicle presence triggers OCR processing
03
Container OCR
Container numbers read in real time
04
Result Selection
Best read selected across frames and camera angles
05
JSON → MQTT
Result published — any subscriber acts on the data
Steps 01–02 — Trigger

Motion-Driven Processing

Tracker monitors all connected camera streams continuously. Processing is triggered by motion detection — OCR runs only when a vehicle is present in the frame, keeping hardware load proportional to actual traffic volume.

Steps 03–04 — Read

OCR & Best-Result Selection

The ConPDS Core OCR engine reads validated container numbers — including BIC-prefixed and ISO Tank formats — across multiple frames and camera angles. Tracker selects the highest-confidence result — up to 99% accuracy at 30ms processing time.

Step 05 — Output

Open JSON Output via MQTT

Every confirmed read is published as a JSON message to the MQTT broker — container number, camera ID, camera group, timestamp, and confidence score. Any subscribed system receives it instantly: your DMS, a custom integration, a database writer, or ConPDS Autogate.

Three Capabilities That Make Tracker Infrastructure, Not a Product

Tracker does not prescribe what happens after a container number is read. It produces a container-linked evidence record — accurate, structured, on-premise — and leaves the business logic entirely to the operation that owns the data.

Capability 01 — Identification

Container Number Recognition

Reads validated container numbers — including BIC-prefixed and ISO Tank formats — horizontal and vertical orientation, single- and double-line
ISO container type recognition: size and type codes validated against the container number standard
Rear camera speed up to 10kph · side camera up to 7kph · angles up to 30° horizontal and vertical
Capability 02 — Output

Open JSON Output via MQTT — Build Any Integration

Every read published as structured JSON to the MQTT broker — no proprietary format, no vendor lock-in
Integration via MQTT, REST APIs, or direct database integration — standard protocols, no middleware required
See the full Integration Guide →
Capability 03 — Infrastructure

On-Premise, Any Camera, Any Facility Size

Runs on your hardware — Windows 10+ / Linux Ubuntu 20+ — nothing leaves your network
GPU-accelerated processing scales with hardware as camera count grows
See full Technical Specifications →

What Container Operations Gain With Tracker Infrastructure

track_changes

Up to 99% OCR Accuracy

Deep-learning ConPDS Core OCR engine with motion and object detection — container numbers read correctly the first time, every time a vehicle passes.

flash_on

30ms Processing Time

From camera frame to JSON result in as little as 30ms — downstream systems receive data before the truck has cleared the lane.

photo_camera

Works With Existing Cameras

Any IP camera supporting RTSP streams — no proprietary hardware required, no ripping out existing infrastructure. Axis, Hikvision, Bosch, Vivotek and others.

lock

No Management System Dependency

Works with any DMS, TOS, or WMS — or none at all. Integrate via MQTT, REST APIs, or direct database integration. You choose what consumes the data.

home

Fully On-Premise

All processing and data storage happens on your hardware. Nothing is sent to ConPDS systems. You own the data and control retention, access, and compliance.

sync

Centralised OCR Engine Updates

Update Tracker once — every connected camera stream benefits instantly. No manual firmware upload per camera, no per-device reconfiguration. Cameras with built-in OCR require individual updates; Tracker doesn't.

visibility

Motion-Triggered Processing

OCR runs only when a vehicle is detected — not continuously across all streams. GPU load stays proportional to actual traffic, reducing hardware requirements and energy use at facilities with variable throughput.

settings_input_antenna

Concurrent Multi-Stream Processing

One Tracker installation processes multiple camera streams simultaneously — 5, 20, or 50+ cameras handled by a single OCR engine. No separate system per camera, no per-stream licensing complexity.

square_foot

Scales to Any Facility

From a single gate camera at a small depot to multi-lane, multi-site terminal deployments — same software platform, same open output format, same central update model throughout.

What Do You Want to Build on Tracker's OCR Output?

Tracker publishes container numbers as open JSON via MQTT — what happens next is your choice. Two paths: build your own integration on the Open Platform, or deploy ConPDS Autogate for full gate automation at DepotMaster depots.

Open Platform · Any DMS · Any Workflow

Build Your Own Integration

Subscribe any system to the MQTT feed — your DMS, TOS, WMS, a custom webhook, or a direct database write. No ConPDS middleware required.

Pre-Built Solution · DepotMaster Depots Only

ConPDS Autogate — Ready to Deploy

Complete gate automation built on Tracker — cameras, kiosk totem, barrier, intercom, and full DepotMaster integration. Tracker's OCR output becomes barrier decisions and self-service driver interaction.

See ConPDS Autogate →

The ConPDS Evidence Platform: Tracker · Checker · Autogate

Tracker is the OCR infrastructure — the evidence source for every container that passes a camera. Checker is the mobile evidence layer for inspection, repair, and lease return. Autogate is the complete gate installation built on Tracker for DepotMaster depots. All three write to the same container-linked record chain — one container number, one operational evidence record, across what the camera sees, what the technician captures, and what the office needs to govern.

Tracker is the evidence source. Checker is the evidence layer for the work the camera cannot see. Autogate closes the loop at the physical gate. Each product works independently — and together they cover the full container lifecycle on a single container-linked record chain.

Architecture, Data Ownership & Integration

Edge / On-Premise Processing No Cloud Dependency Customer Data Ownership Open JSON / MQTT Output Any IP Camera Compatible No Management System Dependency Windows & Linux Support
All processing runs on-premise on your GPU hardware — container numbers and video frames never leave your network
You own the data ConPDS Tracker produces — data retention, access controls, and GDPR compliance are your responsibility and under your control
OCR results published via MQTT in standard JSON — integrates with any DMS, TOS, WMS, or logistics platform

Technical Specifications

Performance Highlights
Accuracy
Up to 99%
Processing time
As low as 30ms
Standards
ISO 6346 · BIC · ISO Tank
Camera protocol
RTSP / RTSP-S
Platform Requirements
OS
Windows 10/2016+ · Linux Ubuntu/Debian 20+
GPU
Nvidia CUDA · Intel HD
RAM
Min. 8GB (16GB+ recommended)
Architecture
On-premise, GPU-accelerated
See the full Technical Specifications — complete OCR performance benchmarks, camera requirements, hardware specifications, authentication methods, and architecture details.

Frequently Asked Questions — ConPDS Tracker

What is ConPDS Tracker and what does it do?
ConPDS Tracker is an on-premise container number OCR engine. It connects simultaneously to multiple IP camera RTSP streams, uses deep-learning recognition to read validated container numbers — including BIC-prefixed and ISO Tank formats — in real time, and publishes results as JSON via MQTT — per camera or per camera group. Any system that can subscribe to an MQTT topic can consume and act on that data: a DMS, TOS, WMS, a custom integration, or a direct database write. No management system dependency, no cloud requirement, no proprietary lock-in.
Does ConPDS Tracker require DepotMaster or any specific depot management system?
No. ConPDS Tracker is completely independent of any depot management system. It publishes container number data in open JSON format via MQTT — any software that can subscribe to MQTT can integrate with it. If your facility runs DepotMaster and you want a fully automated gate solution with barriers and a driver kiosk, ConPDS Autogate is the pre-built integration designed for that scenario.
What camera hardware does ConPDS Tracker work with?
ConPDS Tracker works with any IP camera that outputs an RTSP or RTSP-S stream using H.264, H.265, or HEVC encoding — minimum resolution 1280×960 at 12fps. Cameras from manufacturers including Axis, Hikvision, Bosch, and Vivotek are compatible under these requirements, as is any other vendor whose cameras support standard RTSP output. In most cases, existing on-site cameras are compatible without replacement. ConPDS provides onsite consultation on camera placement, angles, and configuration to ensure optimal read accuracy for your specific installation.
Who owns the data ConPDS Tracker produces?
You do. ConPDS Tracker runs entirely on your hardware on your network — all OCR results, video frames, and movement records stay on your infrastructure. Nothing is sent to ConPDS cloud systems. You control data retention periods, access rights, and GDPR compliance for your installation.
What is the accuracy and speed of ConPDS Tracker's OCR?
ConPDS Tracker achieves up to 99% container number recognition accuracy using the ConPDS Core OCR engine with motion detection and object detection. Processing time is as low as 30ms from frame capture to JSON result. Rear camera reads support vehicles moving up to 10kph; side camera reads support up to 7kph.
Can I build my own integration on top of ConPDS Tracker?
Yes — that is the design intent. Every OCR result is published as standard JSON to an MQTT broker on topics per camera or camera group. Any MQTT client can subscribe and act on that data: trigger emails, write to a database directly, call a REST API, update your DMS, TOS, or WMS, or build a complete gate automation workflow. Integration is also available via REST APIs or direct database integration. For more complex scenarios — custom routing rules, automated alerts, or bespoke workflow logic — ConPDS's team can help design and build that as a separate engagement on top of Tracker's output data.
What is ConPDS Autogate and how does it relate to Tracker?
ConPDS Autogate is a complete automated gate solution built on top of ConPDS Tracker — it adds cameras, a driver-facing kiosk totem with QR scanner, a safety-interlocked barrier, a video intercom, and full DepotMaster integration. Tracker does the OCR and publishes the result; Autogate is the pre-built business logic layer that turns that result into a barrier decision and self-service driver interaction. Autogate is designed specifically for depots running DepotMaster.
Customers across Europe, Africa, Australia & the Middle East
Alisan Den Hartogh Aktas APM Terminals Autamarocchi CCIS Contrepair Cut Coal Eimskip Eurobox Gruppospinelli HRS Isotank Central James Group International Medlog Namops Logistics Porpet QTerminals Kramer Rotterdam Rhespa Europe SDT Marine & Cargo Washmed Werra Kombi Terminal Zuidnatie Zwennis Containers

Start Reading Container Numbers Automatically

Manual container number entry creates errors that compound downstream — wrong numbers in your DMS, mismatched records, and no camera evidence when disputes arise. ConPDS Tracker eliminates manual entry at the source.

✗ Without Automated OCR
Transcription errors enter your DMSA single transposed digit creates a mismatch that may not surface until a dispute is raised weeks later — with no camera record to correct it.
Gate throughput capped by staffEvery movement requires a gate officer to read, type, and verify — throughput ceiling rises and falls with headcount, not container volume.
No timestamped movement recordThe only evidence of a container movement is a manual log entry — which can be wrong, incomplete, or missing entirely.
Data reaches systems late and incorrectlyManual entry introduces lag and transcription errors before data reaches your DMS, TOS, or WMS.
✓ With ConPDS Tracker
OCR reads published at sourceContainer numbers read directly from camera frames — no human transcription in the loop. Up to 99% accuracy, timestamped at the moment of capture.
Gate throughput independent of headcountTracker processes every movement automatically — gate officer attention freed for exception handling, not data entry.
Every read timestamped and camera-attributedContainer number, camera, camera group, and timestamp logged for every movement — unambiguous, unedited, retrievable.
Structured JSON delivered in 30msResults published to MQTT immediately — any subscribed DMS, TOS, WMS, or custom system receives accurate data within 30ms of the read.

Two paths forward — pick the one that fits your operation:

Open Platform

Build Your Own Integration

Deploy Tracker, subscribe to the MQTT feed, and build exactly the business logic your facility needs — any DMS, any workflow, any size.

DepotMaster Depots

ConPDS Autogate — Ready to Deploy

Complete gate automation with kiosk, barrier, and DepotMaster integration — built on Tracker, delivered as one system.

See ConPDS Autogate →
No obligations — reply within 1 business day