This article will describe a project I did as an intern at WSO2.
I was asked to create a system that can visualize the moments of people and identify moment patterns relative to time, as my second intern project. I used this public dataset as my sample dataset.
The system consists of two main subsystems.
- Web application (Front-end) to show data to the end user, request and retrieve data from back-end server
- Back-end server that can evaluate a large text files/ MySql database, process them and expose them as a REST API.
The sample dataset was generated by placing passive infrared motion detectors inside a building(MERL) approximately two meters apart. When a person or group of people walked passing a particular Sensor, it will generate a single data line in the file with starting/ending timestamps and sensor ID. There is a seperate MySql database which maps each sensor ID to its 2D location in a map. ( X and Y co-ordinates). Dataset contains over 30millions activates.
The simplest functionality of the system was to allow users to view Triggered Sensors in a user given time range. As an Example if user wants he can view triggered sensors between 2008 Jan 10, 10.34AM and 2008 Jan 10 ,10.39 AM. If user wants to view triggered sensors in next or previous 5mins (or any defined time range) that also can be done easily as shown in the below video.
Video 1
The next important functionality of the system is that it can play or simulate data set between a defined time range. If I want to simulate what happened in 2008 May 5 08:36 PM, Then it can be done by this system easily as shown below. The key feature of this functionality is that it uses color density of dots to indicate direction of the motion.
Video 2
The system can also used to view the triggered densities of each sensor. As an example in given time range from 2008 May 5 08:36 PM to 2008 May 5 09:36 PM, sensor 214 can have triggered 15 times while sensor 220 have triggered 100 times. This will give us a rough idea on the busyness of a path or area in the selected time range. And also if you click a particular sensor, you can view the frequencies of sensor triggering in upcoming time ranges. This feature is demonstrated below.
Video 3
Typical scenario of a given hour in the area can also be inspect by the system. It will describe the busyness values of each sensor after considering the whole dataset. This feature of the system is demonstrated below.
Video 4
We give data to this system mainly as text files. To upload such text file user interface also can be used. This will take some time as it needs to do some analyzing before save it in the server.
Video 5
Future improvements of the system
- Give user the capability to upload map image and mysql data files through user interface
- Real time data analytics
