Computer vision systems and methods for detecting person dwell time
Example implementations include a method, apparatus and computer-readable medium for computer vision detection of dwell time, comprising receiving a plurality of indications of persons entering an environment over a period of time. The implementations further include adding, into a first-in-first-out (FIFO) queue, a respective identifier of a respective person entering the environment and an entry time for each of the plurality of indications, removing, from the FIFO queue, the respective identifier and the entry time of the respective person in response to an indication of an exit of the respective person from the environment at an exit time, calculating an estimated dwell time of the respective person based on a difference in the entry time and the exit time, and generating an alert based on calculating an average dwell time of the persons entering the environment over the period of time based on each estimated dwell time.
1 . An apparatus for computer vision detection of dwell time, comprising:
one or more memories; and
one or more processors coupled with the one or more memories and configured, individually or in combination, to:
receive a plurality of indications of persons entering an environment over a period of time;
add, into a first-in-first-out (FIFO) queue, a respective identifier of a respective person entering the environment and an entry time for each of the plurality of indications;
remove, from the FIFO queue, the respective identifier and the entry time of the respective person in response to an indication of an exit of the respective person from the environment at an exit time;
calculate an estimated dwell time of the respective person based on a difference in the entry time and the exit time;
calculate an average dwell time of the persons entering the environment over the period of time based on each estimated dwell time; and
generate an alert in response to the average dwell time being less than a threshold dwell time, wherein the alert comprises recommendations to increase the average dwell time.
2 . The apparatus of claim 1 , wherein the recommendations include at least one of physically modifying a layout of the environment, adjusting opening and/or closing times of the environment, or changing a product sold in the environment.
3 . The apparatus of claim 1 , wherein to receive the plurality of indications of the persons entering the environment comprises to:
receive, over the period of time, a plurality of data frames from one or more cameras depicting a view of an entry point of the environment;
detect the respective person in the plurality of data frames; and
determine that the respective person moved through the entry point in the environment.
4 . The apparatus of claim 1 , wherein the one or more processors are further configured, individually or in combination, to:
receive, over the period of time, a plurality of data frames from one or more cameras depicting a view of an exit point of the environment;
detect the respective person in the plurality of data frames;
determine that the respective person moved through the exit point in the environment; and
generate the indication of the exit of the respective person from the environment based on determining that the respective person moved through the exit point in the environment.
5 . The apparatus of claim 1 , wherein the respective identifier omits personally identifiable information of the respective person.
6 . The apparatus of claim 1 , wherein the indication of the exit is for is caused by the exit of a different person that entered after the respective person, but is linked with the respective person instead of the different person based on an ordering within the FIFO queue.
7 . The apparatus of claim 1 , wherein the one or more processors are further configured, individually or in combination, to:
identify at least one residual entry based on a difference between a number of exits from the environment and a number of entries into the environment during the period of time, wherein the at least one residual entry indicates one or more persons that disrupted an ordering of the FIFO queue;
calculate an expected dwell time of the one or more persons associated with the at least one residual entry; and
calculate an updated average dwell time of the persons entering the environment over the period of time based on each estimated dwell time and the expected dwell time.
8 . The apparatus of claim 7 , wherein to calculate the expected dwell time comprises executing a Poisson function.
9 . The apparatus of claim 1 , wherein the one or more processors are further configured, individually or in combination, to:
receive transaction data corresponding to the period of time, wherein the transaction data indicates an average transaction size; and
calculate a conversion size based on the average dwell time and the average transaction size.
10 . The apparatus of claim 9 , wherein the one or more processors are further configured, individually or in combination, to:
generate another alert in response to the conversion size being less than a threshold conversion size.
11 . A method for computer vision detection of dwell time, comprising:
receiving a plurality of indications of persons entering an environment over a period of time;
adding, into a first-in-first-out (FIFO) queue, a respective identifier of a respective person entering the environment and an entry time for each of the plurality of indications;
removing, from the FIFO queue, the respective identifier and the entry time of the respective person in response to an indication of an exit of the respective person from the environment at an exit time;
calculating an estimated dwell time of the respective person based on a difference in the entry time and the exit time;
calculating an average dwell time of the persons entering the environment over the period of time based on each estimated dwell time; and
generating an alert in response to the average dwell time being less than a threshold dwell time, wherein the alert comprises recommendations to increase the average dwell time.
12 . The method of claim 11 , wherein the recommendations include at least one of physically modifying a layout of the environment, adjusting opening and/or closing times of the environment, or changing a product sold in the environment.
13 . The method of claim 11 , wherein receiving the plurality of indications of the persons entering the environment comprises:
receiving, over the period of time, a plurality of data frames from one or more cameras depicting a view of an entry point of the environment;
detecting the respective person in the plurality of data frames; and
determining that the respective person moved through the entry point in the environment.
14 . The method of claim 11 , further comprising:
receiving, over the period of time, a plurality of data frames from one or more cameras depicting a view of an exit point of the environment;
detecting the respective person in the plurality of data frames;
determining that the respective person moved through the exit point in the environment; and
generating the indication of the exit of the respective person from the environment based on determining that the respective person moved through the exit point in the environment.
15 . The method of claim 11 , wherein the respective identifier omits personally identifiable information of the respective person.
16 . The method of claim 11 , wherein the indication of the exit is caused by the exit of a different person that entered after the respective person, but is linked with the respective person instead of the different person based on an ordering within the FIFO queue.
17 . The method of claim 11 , further comprising:
identifying at least one residual entry based on a difference between a number of exits from the environment and a number of entries into the environment during the period of time, wherein the at least one residual entry indicates one or more persons that disrupted an ordering of the FIFO queue;
calculating an expected dwell time of the one or more persons associated with the at least one residual entry; and
calculating an updated average dwell time of the persons entering the environment over the period of time based on each estimated dwell time and the expected dwell time.
18 . The method of claim 17 , wherein calculating the expected dwell time comprises executing a Poisson function.
19 . The method of claim 11 , further comprising:
receiving transaction data corresponding to the period of time, wherein the transaction data indicates an average transaction size; and
calculating a conversion size based on the average dwell time and the average transaction size.
20 . The method of claim 19 , further comprising:
generating another alert in response to the conversion size being less than a threshold conversion size.