1.5.1 #
- Angle input for sensors in degrees: Simplify sensor configuration by entering the respective sensor angles in degrees instead of radians
- Consideration of sensor inversion during configuration: Easier setup of the sensors, as the parameters to be entered during configuration are independent of the installation of the sensors (sensor head above or below)
- Sensor configuration adjustments when tracking is enabled: Optimize the setup of the TrackingMaster by allowing changes to the configurations of the stored sensors while tracking mode is enabled
- Improved visualization of sensors in “Zone Edit”: The current sensor alignment in relation to the central zero point is visible in the “Zone Edit” view
- Configuration of the position of the sensors in the “Zone Edit”: Possibility to adjust the sensor position including the angle directly in the “Zone Edit” view, analogous to the setting up of zones
- Increased error resistance in network communication with sensors: TrackingMaster can better handle minor network problems and, for example, in the event of a delay in network communication, only the most recent sensor readings are taken into account
- Tripling of the evaluated sensor measurements: Deactivation of the median filter integrated in the sensors for an increased measurement density and corresponding adjustment of the general sensor data processing, which enables more reliable detection of people, especially at greater distances
- Expansion of the zoom levels in “Zone Edit”: A larger zoom range is available in the “Zone Edit” view
- Visualization of remaining measurement values for person detection: In the “Zone Edit” view, all measurement points that remain after the pre-filtering for person detection are highlighted separately as magenta points
- Bug fix for tracking zone creation: A bug has been fixed that could occur when multiple tracking zones were created in certain constellations and required the page to be reloaded
- Global consideration of zone priorities: When evaluating the sensor data, the zone priorities are taken into account globally and no longer individually for each sensor
- Improvement in the evaluation of tracking targets: The assignment of detected legs to tracking targets has been fundamentally revised. Optimizations have also been made in the differentiation between active and inactive targets depending on the priority of the lower zones.
- Merging multiple tracking targets: In the rare case that multiple tracking targets are assigned to a person, the new function called “Quick deletion” attempts to merge these targets to prevent unnecessary orphaned targets from timeouting
- Optimization of preset zone timeouts: Preset zone timeouts can be defined more finely than fractions of a second. A minimum timeout has also been set in this context to ensure more stable behavior when deactivating preset zones
- Own median filter for sensor measurements: During preprocessing of the sensor data, an individual median filter was implemented to smooth out measurement inaccuracies, which does not lead to a reduction in the general measurement density. Different object or material properties lead to fluctuations in the sensor measurements. To reduce these fluctuations, smoothing is carried out by calculating the median over measurement value groups.
- Gaussian filter for sensor measurements: As an alternative to the median filter activated by default, two variants of the Gaussian filter have been integrated to smooth out the measurement inaccuracies. Different object or material properties lead to fluctuations in the sensor measurements. To reduce these fluctuations, smoothing is carried out based on a Gaussian distribution. In most situations, the median filter is the better option and should therefore be used preferentially.
- Own particle filter for sensor measurements: Replacement of the integrated particle filter of the sensors, which is associated with a delay in the measured values, with a separate particle filter that selectively delays measured values only when a potential target is newly detected in order to filter out particles such as dust grains
- Filtering extreme values of the sensor measurements: Measurement values that indicate incorrect measurements by the sensors based on their value are excluded from the analysis
- Filtering static object positions: Objects with a fixed position are learned and can therefore be ignored when analyzing the sensor measurements in order to prevent false detections. Due to the fluctuations in the measured values, especially at edges, the learning process also accepts certain tolerances.
- Single activation filter: This new filter is an additional logic for filtering out individual measurement errors from the sensors in areas covered by zones. Significant one-off measurement deviations are ignored if they cannot be associated with object movement.
- Filter Minimum number of measurements for leg clustering: Depending on the distance from targets, a minimum number of correct sensor measurements is required
- Preparation of optimized target tracking using a Kalman filter: The assignment of leg detections to tracking targets should be further improved by predicting the next position of existing targets based on their movement profile. A variant of the Kalman filter is integrated for this prediction, which is deactivated by default due to ongoing optimization. If only one person is tracked, the filter can be used for testing purposes.
- Deactivation of individual filters: A separate configuration page “Filters” has been provided to activate and deactivate the filters
- Detailed configurations of filters and sensor data processing: For individual steps of sensor data processing and the various filters, more than 40 parameters are available on the new “Filters” page for the individual configuration of the TrackingMaster, which allows better adaptation of the behavior to different situations. The parameters are generally limited to sensible value ranges, but should nevertheless only be adjusted with the utmost care.
- Reset function for filter configurations: All modifications to the parameters available on the “Filters” configuration page can be reset to the default values using a reset function
- Optimized speed functions for the zoom of Canon PTZ cameras: The base curves for controlling the zoom on Canon PTZ cameras have been replaced, resulting in faster zoom adaptation from “actual” to “target”
- Consideration of camera inversion during configuration: Easier setup of the PTZ camera as the system takes into account any possible inversion of the camera in the settings in the “PTZ Viewer” view
- Configuration of additional image sources: In addition to the PTZ camera, other external static image sources can be stored in the TrackingMaster
- Configuration of image outputs: Function for the optional definition of available image sinks in the AV system, whereby freely definable keys can be stored in order to be able to address not only the physical but also logical image outputs that are provided with a specific configuration in connection with image mixers (e.g. for various PIP constellations)
- Configuration of equipment scenarios: So-called equipment scenarios can be stored, which provide commands for controlling other devices in the AV system for media control via freely definable keys (e.g. Spot_light_center_brighter, Micro-phone_lectern_mute_off)
- Definition of active image sources for zones: Each preset and tracking zone can be assigned combinations of image sources and sinks in the “Zone Edit” to enable automatic switching between different image inputs and outputs depending on the positions of tracking targets in conjunction with a media control and image mixers
- Definition of active equipment scenarios for zones: Equipment scenarios can be assigned to each preset and tracking zone, which can be used to control devices in the AV system using a media control depending on the current tracking configuration (e.g. unmute the microphone on the lectern as soon as a person is in a defined area of the lectern)
- Endpoint “activetriggers” of the TrackingMaster API: A new API endpoint is available that has been specially optimized for processing control commands from the TrackingMaster using a media control or similar. In addition to the image sources currently being used in conjunction with image sinks and the active equipment scenarios, the interface provides information on the positions of the relevant targets and the current tracking constellation.
- Further functional extensions of the TrackingMaster API: Several small adjustments as well as integration of endpoints for image sources, image outputs and equipment scenarios
- Revision of the TrackingMaster API documentation: bug fixes, explanatory texts optimized, functionalities for testing the API adapted
- TrackingMaster API documentation Availability: Public provision of the Tracking Master API documentation on https://docu.trackingmaster.io/ regardless of the local accessibility of a TrackingMaster, although the “tryout” functionality cannot be used
- Improved display of update information: Alphabetical sorting of update statuses according to individual modules
- Optimization of logging: The logging of the TrackingMaster has been improved and the maximum amount of collected logging data has been increased
- Recording function for sensor readings: Raw sensor readings can be recorded to enable easier analysis and to be able to repeatedly reproduce constellations
- Various performance optimizations: Internal runtime performance optimizations to achieve identical or sometimes even lower runtime and system load despite the new, partly computationally intensive functions
1.5.2 #
- Renaming of “Ghostmode” to “Pause Tracking”: standardization of the name in the web interface and the API.
- TXT API Bug Fix: Fixed a bug that caused an incorrect power status to be displayed briefly during system shutdown.
- Improvement “Zone Edit”: For each sensor, the corresponding measuring points can be hidden or colored as desired.
- Canon N700 Support: Canon N700 cameras are supported and
can be selected during setup. - Bug fix for camera tilt correction: Fixed a camera control bug that could occur if the camera
was not placed in the center of the room and tilt correction was active.
1.5.3 #
- Canon CR-N100 Support: Canon CR-N100 cameras are supported and
can be selected during setup. - Sections of measured values from sensors can be excluded from the evaluation: To
reduce the negative effects of reflections on high-gloss objects,
areas of the measured data can be completely excluded from the evaluation. - Automatic deactivation of “pause tracking” when switching on the system: To avoid errors, when switching on the system, it is ensured that “pause tracking” is inactive so that automatic camera control can take place.
- API extension: The API can be used to query the normalized camera position regardless of the camera model.
