The following sections provide answers to some frequently asked questions.
Live Sense SDK, a multi platform solution powered by AI, enables apps and devices to detect real-time hazards and road signs to help drivers make informed decisions. It can detect objects, on the road and in the surrounding space, in real-time using just the edge processing on the device that can plug into any type of vehicle, regardless of internet connectivity. A complete list of features can be found here - https://developer.here.com/products/live-sense-sdk
Live Sense core functionality for object detection and classification works for any road surface globally.
The latest version recognizes the most common road signs today, including speed limits, stop signs, and hazard signs. For a detailed list, please refer to our developer guide.
Our current SDK version accurately identifies speed limit signs posted in regions where the signs are similar to signs in:
Reach out to our support team if you would like to have Live Sense SDK increase its coverage.
The best way to mount mobile or embedded devices running Live Sense SDK is on the windshield or on the dashboard so that it has a clear and unobstructed view of the road in front of the vehicle. The performance is best when the mount can hold the device still with no vibrations.
Live Sense SDK uses the CPU, GPU, and processor to run the AI models and to process the detections. We recommend you plug in your device if you plan to use it for long periods of time.
The performance of the SDK depends on how good the lighting is and how clear the sign is. Even our eyes cannot recognize dark objects. The SDK does function in low-light or at night provided the object or sign is illuminated well.
Contact access support if your license key has expired.
The confidence is defined as the accuracy with which a certain object or sign is recognized by a machine learning model. Low confidence values might lead to an increase in false positives when the model recognizes incorrect objects.
A very high confidence value means the model will recognize objects or signs almost identical to those used for its training, and it might ignore a few similar kinds. We recommend a minimum confidence value of 0.6 (60%) for our models.
A detailed list of each model and its recommended confidence value can be found in Models.
For the complete guide to HERE licensing, see https://developer.here.com/faqs#licensing--terms.