Vivotek took advantage of its extensive experience in the surveillance field and vast image databases to develop a neural network-powered detection engine for human forms.

Human Detection

Based on a human silhouette database and rapid responses via a pre-trained deep learning model, the smart engine instantly recognizes the appearance of human forms in the video surveillance area. As humans are the objects of interest in the majority of video surveillance, the Human detection feature enables users to quickly configure installation. With Human detection, only human activities will serve as event triggers, while swaying vegetation, vehicles passing by, or animal activities in the scene will no longer trigger such events. This helps reduce both false alarms as well as the time and effort taken for camera configuration.

This pair of images reveal the enhanced precision of Smart Motion Detection. With red tags indicating alarm, in the left image we can see the numerous false alarms created through traditional motion detection. On the right the same scene is displayed with Smart Motion Detection – only the human figure triggers the alarm.

Robust for Any Lighting Condition

Built from a database of thousands of scenarios in day, night, and various lighting conditions, the Smart Motion’s deep learning algorithm comes with lighting models that can quickly adapt to the user’s installation. The lighting models in Smart Motion Detection can deal with high contrast sun-lit scenes, indoor lit or unlit environments, false alarms caused by vehicle headlights, or other potential distractions. Further, the embedded lighting models mean that users do not need to fine-tune the motion sensitivity for different application scenarios or at different times of a day.

Time Filter

Vivotek also provides advanced time filtering in its Smart Motion Detection technology. The settings for time filtering include two parameters: Minimum Activity Duration and Activity Merge Interval. If a fast-moving object is not to be considered as an event trigger, users can configure the Minimum Activity Duration to a span of seconds an object must stay in the detection area before an event is triggered. On the other hand, the Activity Merge Interval can be configured with the maximum number of seconds allowed to separate motions triggered by a single object. Close occurrences of motions are considered as constituent parts of a single event, so multiple discontinuous movements can be consolidated into a single alarm, thus reducing the number that need to be screened. This ensures that alarms can be triggered even when minute and discontinuous motions are detected.

Intuitive Configuration

Vivotek's smart motion detection technology improves on conventional configuration approaches that rely on the size of objects in the frame, offering instead an intuitive interface displaying how motion activity changes along a time axis in real time. This allows users to clearly apprehend whether the current configuration can accurately detect motion in the scene.

The activity data collected by the camera is displayed in comprehensive graphs. Users can monitor and compare the historical data acquired through a span of time as well as fine-tune the detection parameters easily though the graphical interface.

For Smart Motion users who chose not to use the Human detection feature, the detection threshold can be changed while observing the activities on screen. The activity chart covers a 60 second time span across the X-axis. The Y-axis indicates the percentage of alarmed pixels in the field of view. In this way, the user can ascertain whether appropriate motion sensitivity has been configured.

Conclusion

Vivotek's Smart Motion Detection enables instant recognition of human activities, effectively reduces false alarms, and avoids the time wasted for security staff to verify triggered but false events. On the initial setup, Smart Motion Detection reduces the amount of time required for installation and the chance of making configuration errors, while minimizing the number of false alarms. Smart Motion Detection readily applies to most surveillance scenes without the need for tedious configuration tasks and adjustment efforts on the site.