Human Footfall count in a chain of show rooms with Hikvision CCTV Cameras, DVR, Windows 7 machine, Internet

Overall Context of the Client

1. The Client has about 100 showrooms and wants to roll out our solution in all showrooms.

2. Most showrooms have single entry/exit excepting a few which have multiple entry/exit.

3. Each showroom is equipped with a minimum of 3 CCTV Cameras and a maximum of 8 CCTV Cameras.

4. All showrooms have the same brand of CCTV Camera – HikVision. Most of the cameras are IP Cameras, but for a few which are old analog cameras.

5. Each showroom has broadband internet connectivity.

6. Each showroom has a Windows 7 based central machine that is used for monitoring CCTV camera output, invoicing and billing. These machines in the individual showrooms are then connected to the central monitoring machine at the Corporate office.

7. The client does not allow any changes to be made to the existing hardware. The client also does not allow any new hardware to be added to his network.

8. Our solution has to be installed on the Windows 7 based central machine in each show room as well as in the central monitoring machine at the Corporate office. The GUI and Monitoring will be at the Central monitoring machine at the Corporate office.

Key requirements of the client

Centralized monitoring of footfall

a. Footfall count of individual walk-in customers visiting their showrooms

b. Footfall count of small group of people visiting their showrooms

c. Footfall count has to be available at individual showroom level as well overall Bengaluru level, but the Monitoring / GUI will happen at the Corporate office. Count will have to be maintained at the show room computer and pushed to the Corporate office system periodically

d. There may be 2 to 6 employees in each show room depending on the size of the show-room. We need to have a way to avoid including them in footfall

SCOPE of this project work: The solution should work in 3 showrooms for 3 weeks with consistent performance with a minimum accuracy of 85%. Transfer of technology to a designated person: Installation procedures, Configuration procedures, Well documented code with comments.

Suggested Approach

1. Track Objects (Persons) for a period of 12 hrs (Usually 10 a.m to 10 p.m, configurable) - capturing the following details :

- Total Count of Objects, Total Count of Objects that spend Less than 2 hrs (Configurable Parameter in mins).

- Capturing groups :

-When any new object is found it should be considered as new group, and all the new objects found within 30 sec

(configurable parameter in sec) next to each other should be added to the same group created for the first object.

-Total number of Groups with more than one object, and List of groups with number of objects which more than one.

2. Tracking should support up to 5 RTSP streams from a DVR. 2 to 5 RTSP streams (CCTV) will be configuable depending on the size of the show room and position of the CCTV.

3. Tracking should be reset for every day. Reset time may be at 00:00 IST.

4. Update all the details to a provided REST API every day - Details will be provided later. Storing and display at the Central monitoring station will not be your responsibility. We will receive via REST API and store in DB and also display on dashboard.

5. Tracking of objects should be unique irrespective of RTSP stream that will be configured. (from configured CCTV sources).

The person should be tracked across RTSP streams coming from various CCTVs.

6. The Complete setup of this module should be dockerized for windows 7 Machine as well as Linux.

7. The Captured details for every day should also be available in a local DB data stored up to max 45 days, in case network fails. REST API to extract data from locab DB should be made available.

OpenCV NumPy Python Docker MQTT

ID do Projeto: #28798766

Sobre o projeto