ObjectTrack Pro

Use ObjectTrack Pro to resolve the use cases:

• Man on a rail track — to take visual control over regular maintenance works on the track, detect intentional or accidental intrusions into restricted areas.
• Abandoned luggage — to promptly reveal potential threats to public safety and human lives.
• Perimeter Intrusion — to safeguard restricted areas, control access, etc. against humans or vehicles.

ObjectTrack Pro in general

How it works

At the core of the module, there is a convolutional neural network CNN-1 trained to detect objects of pre-defined classes. It processes every n-th frame in the incoming video stream to build annotations (bounding boxes) around the detected objects; in between, the objects are tracked.

Then CNN-2 comes into action to find out what specific class the object belongs to (e.g. if it is a human in "Man on track" use case) and pass it for a check against user-defined class rules (e.g. if a tripwire is crossed in "Perimeter intrusion" use case); whenever a rule applies, the module triggers an event.

Tips to use ObjectTrack Pro

This is a last-generation AI module in which highly efficient object recognition algorithms are implemented. It allows ObjectTrack Pro to demonstrate impressive statistical metrics even under unfavorable outdoor conditions (unlike MotionTrack Pro).

Camera requirements

The best performance of ObjectTrack Pro is guaranteed under the following conditions:

• Camerа mount: fixed only (no PTZ, mobile phone uploads, or drones).
• Resolution: from 1280x720 (HD Ready) to QHD 2592x1944.
• Frames per second: 10 - 30 fps.
• Contrast object vs background: 20%.
• Tilt: +15° to +90°
• Pan: -75° to +75°
• Roll -5° to +5°
• Image quality within the region of interest: over 50 px/m
• Height: from 3 to 15 meters above the ground level.
• Minimum detected object size:
  • 1024 (32Х32) px for abandoned bags and humans
  • 2025 (45Х45) px for vehicles

Computational complexity

ObjectTrack Pro is rated "High"; the exact numbers may vary with video resolution and CPU parameters.

Export module settings

Export the module settings to a *.xml file, e.g., to pass it to qualified IREX personnel to find a solution to a problem with the module performance, should it happen. To do so:

1. Find the camera running the module in the User menu — Settings — Cameras, and click Video  analytics settings as shown:    

2. On the VIDEO ANALYTICS tab on the camera settings pane on the left, click the three-dot menu 



Export analytics configurations of the module — to export the module settings on the MAIN sub-tab.

Export analytics configurations of the model — to export the entire module configuration, including the rules, if any.

Enable ObjectTrack Pro

1. Find the camera to enable the module in the User menu — Settings — Cameras section and click Video analytics settings as shown.    

2. On the <module_name> list, click the name of the module.
3. Slide right the On/Off slider 
4. Click Apply at the bottom of the pane.

Detect men on a rail track

The use case, for sure, is a real-time one, since it obviously needs quick reaction. So:

1. Make sure the cameras tasked with rail track surveillance meet the system requirements.
2. Set up ObjectTrack Pro on those cameras to detect men on a rail track.
3. Build an alarm monitor with Cameras = those in item 1 and Analytics modules - ObjectTrack Pro - Motion in region checkmarked.
4. Enable notifications about new alarms on this monitor.

1. On the MAIN sub-tab, define the parameters:

Region of interest - the part of the frame to analyze is shown bordered in green. Defaults to the full frame.

To move a vertex, point the cursor onto the vertex and drag and drop it.
To delete a vertex, double-click it.
To add a vertex onto an edge, double-click where the vertex should be.
To move the region as a whole, place the cursor inside it and drag-and-drop.

View: select Top or Side.
Object type: switch on People.
Ignore people on trains: switch on not to consider people inside the train box "Persons on track", e.g., like this:



Other settings on this tab are irrelevant to the "Man on track" use case.

2. On the RULES sub-tab, click Add rule to select the Motion in region on the open menu.

3. On the Motion in region card that opens, enter a custom name of the rule (e.g., "Man on track").

It makes sense to give a rule a self-explanatory and meaningful name since it will be the event name that the users are meant to operate with. Rule names can be edited on the User Menu — Settings — Event Names screen.

4. Click the Region of interest under CALIBRATION to build the region where to detect humans right on the rail track.

5. Click Apply.

Find abandoned luggage

Unattended bags, suitcases, and other items may pose significant danger at transportation hubs and other public spaces. So, to be informed in real time about incidents of this kind:

1. Make sure the cameras tasked with rail track surveillance meet the system requirements.
2. Set up ObjectTrack Pro on those cameras to find unattended items.
3. Build an alarm monitor with Cameras = those in item 1 and Analytics modules — ObjectTrack Pro — Abandoned item checkmarked.
4. Enable notifications about new alarms on this monitor.

1. On the MAIN sub-tab, define the parameters:

Region of interest - the part of the frame to analyze is shown bordered in green. Defaults to the full frame.

To move a vertex, point the cursor onto the vertex and drag and drop it.
To delete a vertex, double-click it.
To add a vertex onto an edge, double-click where the vertex should be.
To move the region as a whole, place the cursor inside it and drag-and-drop.

View: select Top or Side.
Object type: switch on the Abandoned bag.
Consider abandoned anew after ... min: one "Abandoned item" event duration at the same location.
Ignore bags closer than: if switched on, a bag is not considered abandoned if a person is close to it. The "close" size is defined as the Bag filtration area enlargement factor x 100% x bounding box size. In the example below Bag filtration area enlargement factor = 2, so the "not abandoned" area (red box) is twice as big as the bag bounding box (yellow box).



Other settings on this tab are irrelevant to the "Abandoned bag" use case.

2. On the RULES sub-tab, click Add rule to select  Abandoned bag on the menu that opens.

3. On the Abandoned bag card that opens enter a custom name of the rule (e.g., "Abandoned luggage").

It makes sense to give a rule a self-explanatory and meaningful name since it will be the event name that the Platform users are meant to operate with. Rule names can be edited on the User Menu — Settings — Event Names screen.

4. Click the Region of interest under CALIBRATION to build the region in which to search for an abandoned item, e.g., like this:

5. Click Apply.

Detect perimeter intrusion

Use the "Line crossing" or/and "Region entrance" events by ObjectTrack Pro to be informed about intrusion attempts into a protected perimeter by humans or/and vehicles. Do so in a way much like that with MotionTrack Pro, although ObjectTrak Pro performs better due to its advanced calculation algorithms and design.

1. On the MAIN sub-tab, define the parameters:

Region of interest - the part of the frame to analyze is shown bordered in green. Defaults to the full frame.

To move a vertex, point the cursor onto the vertex and drag and drop it.
To delete a vertex, double-click it.
To add a vertex onto an edge, double-click where the vertex should be.
To move the region as a whole, place the cursor inside it and drag-and-drop.

View: select Top or Side.
Object type: switch on the People or Vehicles or both depending on the kind of intrusion expected.
Vehicle type and color recognition: recognizes vehicle type and color; enabled at Object type =Vehicles.
Save tracks: stores the motion trajectories of vehicles and humans for later search in the media player.

Other settings are irrelevant to this scenario.

2. Click the Add rule on the RULES sub-tab to select Line crossing on the open menu.

3. On the Line crossing card that opens, enter a custom name of the rule (e.g., "Perimeter intrusion").

It makes sense to give a rule a self-explanatory and meaningfull name since it will be the event name  that the Platform users are meant to operate with. Rule names can be edited on the User Menu — Settings — Event Names screen.



4. Draw the signal line (line not to cross) by drag-and-dropping the default orange shape right on the frame.

The tripwire shouldn't be placed near the screen's border. The minimum distance from the tripwire's points to the screen's border shouldn't be less than half of the object's size.

5. If both People and Vehicle were checkmarked at step 1 above, select which of them to trigger a "line crossing" event on the Filter objects by type list.
6. Click Apply.
7. On the MAIN tab, checkmark Region entrance and define other "Region entrance" event settings, then click Apply.

The "Off-duty" use case

Run this use case to monitor the continuous presence of, e.g., security men at the checkpoints. To do so:

1. On the MAIN sub-tab, checkmark People under Object type.
2. Click the Add rule on the RULES sub-tab to select "Off-duty" on the open menu.
3. On the event card that opens, give a name to the "Off-duty" event.

4. Enter the allowed absence time limit into the "Off-duty" time limit box.
5. Click under the CALIBRATION to draw the mandatory presence area.
6. Click Apply.

Set up right-of-way violations

ObjectTrack Pro takes part in the detection of right-of-way traffic violations. In this use case, its function is to detect the presence of pedestrians and their motion directions in "Crosswalk" areas, which are built in the module settings.

1. On the Main tab, checkmark People-Pedestrians to enable the Traffic objects tab.




2. Click "+" on the Traffic objects tab to open the Crosswalk card.




3. Enter the name of the crosswalk (or of a crosswalk segment).
4. Build the border of the area in which the presence of pedestrians will be controlled.
5. Show the pedestrian motion direction that will be considered "Forward".
6. Click Apply.

Set up safety gear detection

Configure this module to detect safety gear, such as helmets and safety vests, on individuals at construction sites or other high-risk locations. To configure detection:

1. Select People under Object type.
2. Enable Safety gear detection under Extra options.




3. On the RULES sub-tab, click Add rule and select Motion in region.
4. In the configuration panel, define the region where safety gear compliance is required, then click Apply.
5. Enter a descriptive event name (for example, Safety gear control).

With this configuration, the module performs the following:

1. Generates a Motion in region event (using the name defined in step 5) whenever a person is detected moving within the specified area.
2. Evaluates whether the detected person is wearing the required safety gear and records the result as an event attribute (for example, Helmet, no vest or No helmet, no vest).

The detected attributes are displayed in the event details when reviewing safety rule violations.

Human silhouette detection

Switch on the Detect human silhouette under Extra options to run the calculations of the silhouette descriptor for every human found in the frame. This allows searching for a person of interest based on his or her silhouette.