When adding a new camera to ConPDS Tracker, some mandatory settings must be configured, and others are optional.

The Camera Connection screen looks like this.

Camera Connection

All fields are mandatory, and the camera stream is connected if connected correctly.

Camera features

Once a camera is configured, there are different features available.

OCR

Will do OCR of BIC and/or ILU codes. If not enabled, no OCR is done.

Container Detector

This is a data model developed to look for container-like objects. Once an object is detected, OCR processing is initiated.

If disabled, it is possible to draw ROI (Region Of Interest) on the Camera preview screen. OCR processing is done only within the ROI area. Everything outside is ignored.

Motion Detection

The Motion Detection data model keeps the Container Detection and OCR idle to minimise CPU and GPU processing. If a motion is detected, then Container Detector is started.

OCR settings

If OCR is enabled, there are some settings available.

Camera Vote count

Influences on results in a Camera Group. Has a range from 1 to 10.

OCR Minimum confidence

Set up minimum confidence in the Camera Group group result. A confidence of 84 usually means a 96+ per cent accurate OCR result. If the OCR confidence is lower than configured, it is not included in the Camera Group result.

Use Global setting

Forces to use global confidence setting. If enabled, the camera-specific settings are ignored.

Optimization of OCR detections

Use 180-degree rotation

If enabled, Tracker will do OCR of normal view and OCR of 180-degree rotated view if the origin doesn't return any result.

It should be enabled if the camera points to the top of the containers.

Recognition width

Allows ignoring containers in specified recognition width where they move. Enabling this feature depends on the Container Detector model being enabled.

The Recognition width has a range of 1 to 5. If the max range is set, the Tracker will do OCR on all founded containers.

ROI optimization

Predicts the location of the container code in the current frame based on the previous result

Enable this if the camera is pointing to the side of the containers.

Ignore noise after processing

Removes rects that don't follow the BIC container code pattern.

Example of using Recognition width:

In the example below, we see containers in two lanes:

A) Container which needs to be recognised

B) Container which needs to be ignored

C) Indicates the approximate width of recognition where Tracker will do OCR.

If we set the Recognition width range to 2.5-5 for this example, container B will be ignored, but A will not. 

If we set the range to 1- 2.5, container A will be ignored, and container B will not.

Camera MQTT topic

Tracker will publish all OCR results under this topic if an MQTT broker is configured and enabled.

ConPDS Checker Integration

This feature integrates ConPDS Tracker with ConPDS Checker. The picture of the recognized container code is automatically uploaded to ConPDS Checker whenever a container code is detected.

ConPDS Checker Integration

Enables the Checker Integration functionality, depending on "Connected Server".

Before enabling this, the Checker ID should be authorized on a server where ConPDS Checker REST APIs are installed.

Connected Server

Setups the Checker Integration server. There are two predefined and custom options.

Example:

In the example below, the feature is enabled, and we see a connection to the ConPDS Checker server for the customer named "ConPDS".

Pictures of the Truck plate and the container codes are now uploaded to the ConPDS Checker server for searching and previewing.