Object detection
CREATE OBJECT DETECTION BASED ASSISTANCE IN YOUR APPLICATION
Using the object detection module, developers can more easily integrate advanced object detection features. For example, this feature can increase the safety awareness for a machine operator by highlighting obstacles or hazards in the video stream.
The module includes prebuilt object detection demos and examples for how to implement a solution, as well as the runtime and SDK needed to start development and deploy your own application on a CCpilot display.
The object detection module is supported by all our iMX8 based displays, CCpilot V700, V1000 and V1200 equipped with the Google Coral Edge TPU™ accelerator. The best performance is achieved by using the integrated Mini PCIe version.
On the V700 the external USB version of the accelerator can be used. This is recommended mainly for test and evaluation purposes.
The Google Coral accelerator uses the Tensorflow Lite library as its AI
engine and the trained model must be in the tflite format. The module contains an example showing how a model is used with the accelerator and how to handle the result and present it to the user. Several types of detections are supported, for example object detection, pose detection and image classification. By using the accelerator for the AI inference, there is plenty of processing power available for the rest of the user experience and other applications.
Supported products:
- CCpilot V700 – with attached USB AI accelerator
- CCpilot V1000 – with built-in mPCIe AI accelerator
- CCpilot V1200 – with built-in mPCIe AI accelerator