Analytics project: Shopping malls and in-doors commercial centers

Many DISPL partners and end-customers apply DISPL's analytics collection solution in big commercial centers and shopping malls to better understand foot traffic and visitor demographic information, as well as the customer journey.

Let's prepare for this project together:
 

Shopping Mall general project profile

Action takeaways for this project profile

1. Visitor density: Medium-high
   a) Place one camera every 16 m² in areas of interest. The more cameras, fewer contact losses. (Ex: Entrance 1 is 10m wide, first 2 meters length are covered by one camera, next 2 meters require one more camera.)
   b) Point cameras to visitors' major flux direction to capture more faces, avoiding undesired sideways capture.
   
   Full explanation:
   It is expected that multiple people walk together in groups through this environment. Camera positioning should be planned considering the reach of all people face-forward, to avoid losing people that may hide behind their partners. Visitor may block each other's visibility.
   Entrances, such as (1) above, should be covered having the walk direction in mind.
    

   ---

2. Behavior expectations: Fairly unpredictable
   a) Use more than one camera per area of interest, to avoid contacts losses due to unpredictable behaviour.
   b) Avoid face detection in waiting areas.
   
   Full explanation:
   People may change directions, come and go several times, talking to each other, express several emotions in a single contact. Also, sitting, waiting, idling may interfere and make the analytics collection more challenging.
    

   ---

3. Dwell time: Long
   a) Set a fitting session time, to avoid closing a visitor's session prematurely. Cases may come where a customer is identified at the entrance, then disappears inside a store or food court, emerging 3 hours later. This visitor must not be double-counted as new customer, evidently. So, session time should be, at least, double this amount of time.
   
   Full explanation:
   Visits may be as long as 2 hours, maybe more. So, we should prepare for cases when the customer leaves the detection zone for long periods of time.
    

   ---

4. Detection range: Far
   a) Make sure P2P synchronization is set and all devices communicate with each other, to avoid double-counting a customer.
   b) Lessen the minimum face size, in case your camera doesn't detect people at a 4-meter range for any reason. Please exercise caution, for lessening this parameter may cause less reliable detections at long range.
   
   Full explanation:
   Shopping malls are very big areas generally, wall-to-wall distance easily up to 25 meters. Preparations need to take into account the challenge of distance.
    

   ---

5. Lighting conditions: Great
   No actions to be taken! Lighting in shopping malls are ideal.