Video Insights is helping you to speed up your traffic analysis thanks to various object properties collected during the video processing (more described in this article). Based on these properties, we defined a set of metrics and visual maps and their outcomes are presented to you in a form of a Widget. This article will provide you a quick reference to all widget types available in the Video Insights application including their limitations and usage examples.

Red section highlights the widget types section in the New Widget tab

METRICS

1. Total volume

Total volume represents a total number of all traffic objects of selected classes that fulfill the selected filter (movement, event or scenario) in the selected date time range. It can be displayed as one total number or as a pie-graph with counts for particular vehicle types (incl. pedestrians and cyclists). Total volume can be computed for all filter types and their combinations (any defined movements, events or scenarios) or without a filter (to obtain a general overview of the vehicle counts). 

Examples: total number of vehicles crossing a defined line, total number of pedestrians inside a specified zone, total number of vehicles following a given movement, event or scenario.

Total volume widget presented as one number / a pie graph

2. Average volume

Average number of vehicles that meet a defined filter (movement, stay in a zone etc.) for a certain period of time. The granularity can be set from 1 second to 24 hours (for more information about time units please see this article). Visualization of the data is either a total count or a pie graph for the particular vehicle types. Average volume is computed as total volume divided by the specified time unit.

Examples: average number of cars crossing a specific line per minute, average number of vehicles staying in a specified area longer than 30 seconds each hour.

Average volume widget presented as one number / a pie graph

3. Peak Periods

Time periods with heaviest traffic identified during the video footage. Peaks can be shown for one selected filter or for all objects on the scene, however, they are not available for more than one filter (e.g. 2 for movements at once).

Video Insights provide 2 ways for determining peak periods:
a) Peak hour shows the time frame of 1 hour per day with the heaviest traffic for the desired area/movement.
b) AM, PM peaks represent 2 hours with the heaviest traffic per day, one for AM hours  (0-12 o'clock) and one for PM hours (12-24 o'clock).  

Peak interval is normally one hour long and is floating, as a standard peak period is calculated as a four consecutive 15min-long periods with highest sum of traffic.
Therefore, this metric does not have much sense for videos shorter than 1 hour by which actually the same results as by the total volume are reported. If the videos are longer than 24 hours, the peak periods are reported for each day (or part of the day which the video includes).

Example:
Imagine a 12 hours long traffic video lasting from 6AM to 6PM for which you would like to see the peaks. The Peak hour widget would show you that e.g. heaviest traffic in your analyzed material was between 11:20 - 12:20 (floating hour) whereas the AM, PM peaks widget would indicate that the AM peak hour is 11-12 o'clock and the PM peak 4-5 o'clock.   

AM/PM peaks in case of a 32 hours long video / peak hour of a 3 weeks long video

4. Object distribution

Object distribution graph shows the number and type of all traffic objects present in the specified zone/movement during a specified time. The granularity can be set from 1 second to 24 hours (for more information about time units please see this article), and the graph can be generated as a line graph, bar graph or a heatmap.  

Example: You already know what the total number of vehicles was in your 5 hours long scene, but you want to see the distribution of the objects e.g. in 15 minutes long intervals. By doing that, you might find out that, for example, the traffic was the heaviest in the first 30 minutes of the footage. 

Interactive line graph for object distribution with 1 hour time unit

5. Occupancy time (Mean Occupancy Time in Zone)

An average time that the filtered traffic objects spent inside a zone which is selected within a stay event or a movement crossing a zone. It can be created for a movement through a zone and a line/s as well - in that case, however, the mean occupancy time is computed only for the zone.
As all objects crossing the zone are counted regardless of the their direction (even the ones that end up inside the zone without leaving it or the ones repeatedly entering and leaving the selected zone are counted), mean occupancy time is a good indicator about the density of the traffic in the desired area. But if your interest lies more on the time spent crossing the zone, it is advisable to use another metric - travel time (see below).  
 
Examples: Average time vehicles are staying in a short-time parking place or occupying a crossroad entrance, average time spent on a roundabout etc.

Definition of a parking zone / results for mean occupancy time in zone

6. Travel time (Mean travel time)

Average time that the analyzed traffic objects spent within specific space boundaries. Unlike the Occupancy time, Travel time can be generated not just for a single zone but as well for a movement composed of crossing over two lines or zones (see images below). This metric is used to analyze traffic fluency. It can be shown as a single number or a bar graph based on a specified time unit (for more information about time units please see this article).
For calculation of travel time, exact entry and exit point has to be in place and the vehicle trajectory has to be uninterrupted. Moreover, Travel time widget is computed as median, not mean, so that the extreme values (e. g. parked cars in a zone) do not bias the results.  
There are several possibilities for analyzing travel time:

One zone

Time which the object has spent within the zone, from entering until leaving the zone

More zones

Time which the object has spent between entering the first zone and leaving the last zone, whatever the trajectory between them was

More lines

Time which the object has spent between crossing the first and the last line, whatever the trajectory between them was

Line to Zone/Zone to Line

Accordingly to previous situations, travel time here is the time which objects spent between crossing the first line and leaving the last zone or vice versa.

Examples: Mean travel time brings answers to questions like: How much time in average do the vehicles need to go from a defined point or zone A to point B? How long does it take to cross the bridge for cars, buses, bicycles and how much for pedestrians? What is the mean time the drivers need to make a turn on a crossroad? How much time do they need on various parts of the day?

5min periods on Travel time bar graph / Mean travel time of the whole 3 hours long video  

7. Time gap (Mean time gap)

Time gap is constructed as the time duration between two consecutive traffic objects that appeared on the same spot, crossed the same line etc. The shorter the time gap, the higher the frequency of the traffic in the analyzed spot. Time gaps are applicable for one movement at a time and it is advisable to use it for a single line or zone, as only the first detection spot is taken into account (i.e. when a movement goes through several zones or lines, only the first one is taken for the time gap computation).

One zone

Time difference between the car A and the car B entering the defined zone

One line

Time difference between the car A and the car B crossing the defined line

Two zones

Time difference between the car A and the car B entering the defined zone 1. The zone 2 is ignored for the time gap

The widget in Video Insights shows the median time gap for all selected vehicles in the analyzed place and time either as a single value or as a bar graph for a given time unit (for more information about time units please see this article). However, if you are interested in all individual vehicle time gaps, don't worry, they are available in our .xls reports or you can create the Vehicle list widget for that.

Examples: How busy is a street in the morning, around noon and in the evening? What is the average timegap of the vehicles in and out of the rush hour? Do the drivers respect a safe distance in front of a dangerous place on the road?

Advice: Timegap in a zone or on a line is calculated based on an object crossing in any direction. If you' re analyzing timegaps, be careful where you place the zone or line and check that it corresponds with the desired direction of the trajectories. 

Mean time gap presented as a bar graph with 5min time unit / average time gap on the tracked spot 

8. Vehicle list

List of all traffic objects detected in the selected spot/area during the desired time frame together with their exact entry times (either when a vehicle crossed a defined line or entered a defined zone). The Vehicle list widget might be a bit impractical for a large sets of data. In case the first look at the structure via the application widget is not sufficient for your needs, kindly get the whole intrusion traffic report (vehicle list) delivered on your e-mail.  

Examples: What was the sequence of vehicles on an analyzed spot during 1 hour of the footage? At what time has a car, motorbike or a van entered a pedestrian zone? 

List of all vehicles following a given movement, their object classes and the exact entry times 

Note: When the Vehicle list widget is used for a zone, each contact of an object with the zone is counted as new. That's why the result for total count might slightly differ from the Total volume number.

VISUAL MAPS

1. Trajectories

On the preview image on the Describe scene page, only preview of trajectories is displayed, so that the scene is easily readable for placing zones and lines (more information to this topic here).  
This widget makes it possible to see all trajectories of all objects in the video. By using filters, you can have a look on trajectories only for a selected movement or just for some object classes. This might be useful on rich and more complicated scenes, where it is sometimes difficult to get a quick overview.

Example of pedestrian trajectories crossing a city street 

2. Acceleration heatmap

A great tool to discover places with accelerating and decelerating traffic. Hot areas (red color) represent places where traffic usually decelerates, cold areas (blue color) represent places where traffic usually accelerates.
Acceleration heatmap is calculated only from traffic objects which match your widget settings.
Note: On scenes with significant perspective, places far from the camera might produce inaccurate results. Ideal situation for measuring acceleration is from above.  

Acceleration heatmap shows decelerating areas in front of the crossings 

3. Motion heatmap

A great tool to discover places with heavy or low traffic. Hot areas (red color) represent places with heavier traffic, cold areas (blue color) represent places with lower traffic. Motion heatmap is calculated only from traffic objects which match your widget settings.

Motion heatmap identifies zones with heavy and low traffic on first sight 

4. Hold-up heatmap (Occupancy time heatmap)

A great tool to discover an average occupancy time of vehicles or pedestrians on the whole scene or in specific locations. For Hold-up heatmap, the scene is divided into 10 x 10 grid and for each of the fields, mean occupancy time is calculated based on your selected filter and chosen vehicle classes. Unlike by the Occupancy time widget, the results on the heatmap are shown not just for the specified zone but for all fields on the scene, where the tracked objects appeared.

Red areas represent places that traffic objects occupy more than 10 s.
Green areas represent places that traffic objects occupy between 2 and 10 s.
Blue areas represent places that traffic objects occupy less than 2 s.

Note: Hold-up heatmap is calculated only from traffic objects which match in selected time range. On scenes with significant perspective, places far from the camera might produce inaccurate results.

Mean time spent on entrances to a crossing on the Hold-up heatmap (in sec)  

Did this answer your question?