features

Object Detection

Runs a state-of-the-art deep learning model on GPU for fast and accurate object detection

Flexible

Supports multiple parallel streams and any IP CAM.

Configure model detection parameters and start detection in real-time

Integrated

Interacts seamlessly with our Smart IIoT Controller so you can build intelligent applications applying object detection.

Object Detection

Runs a state-of-the-art deep learning model on GPU for fast and accurate object detection

Runs on our AI edge device, the IIoT AI Controller

Train any type of object or combination of objects for object detection in the cloud and apply inference in real-time on the edge.

Parallel streams /
any IP CAM

Connect multiple IP cameras to the IIoT AI Controller, configure IP camera settings and model detection parameters and start detection in real-time.

Up to 8 camera streams can be served in combination with our IIoT AI Controller

Dynamic

With IIoT AI CAM object detection classes are not carved in stone.

It dynamically detects the classes required for object detection and enables you to decide which classes you actually want to detect objects for.

Configure focus areas for object detection, confidence thresholds etc.

Smart
and integrated

IIoT AI CAM treats a camera as a sensor and hence provides all detected objects and counts of objects as measurements in real-time.

Which classes you want the model to detect is defined by how it is trained in the cloud. The final model definition is served to IIoT CAM by our Smart IIoT Controller's OTA service

real-time Object Detection

IIoT AI CAM turns object detection to IIoT as continuous sensor measurements

Training is done in the cloud through AWS DL AMI and results are tracked in the cloud
Generated model weights are automatically stored
Model weights can be applied to the object detection micro-service configuration
The OTA service or our Smart IIoT Controller deploys the weights to the model service
The model service runs the inferencing

The object detection algorithm of IIoT AI CAM is based on a state-of-the-art CNN network (YOLOV5) which is as well accurate as fast.

Training is done by using labelled images (supervised learning) and is done in the cloud on AWS DL AMI.

Results are tracked and based on the results model weights are selected and applied to the model service configuration.

Our Smart IIoT Controller's OTA service will make sure that the correct weights are used for the model service on the edge.

Use the IIoT AI CAM app to configure the input streams and output mjpeg streams of connected cameras.

Change detection regions and line crossing areas for each camera stream.

Change the actual IoU and Confidence thresholds and the actual object classes to detect.

Start detection and track detected objects with live camera streams and/or measurements in the cloud.

IIoT AI CAM can be used to combine with other sensor measurements.

Potential applications are: detection of amount of people in a room or particular area, control of safety gear of people (e.g. helmet, vest), etc.

How it works

Download the

Brochure

Book a

Demo