Pedestrian Crossing Analytical Count and be able to detect the signal for pedestrians.

Pedestrian Crossing Analytical Count and be able to detect the signal for pedestrians. Problem to solve: The program should be able to count each person individual regardless when people are walking very close together, facial detection should be the priority and will be showing in a green rectangles, The program also have to look at the pedestrian signal light if DON’T WALK is on now make a separate count on people still crossing inside the mark regions on the screen. The program will have a setting mode for the End User to Draw the Walking Regions in 4 step points, then adding two lines to mark the Entry and Exit where the direction of the person must be taken in account of entry and exit time stamp of each person. In the setting End User should have the ability to mark the area where the WALK / DON’T WALK sign is in relation to the video screen marking the region with 4-point polygon. Training data must be used to the detection of objects (People and Face). Please Blue Rectangle for the People Object and Green Rectangle people faces. Keep in mind that WALK / DON’T WALL signal has different ways to be display. The End User should have the ability to change in the settings: To Record or Not in an interval time to the Database, SQLite is OK. On this setting also the End User can change To Save or Not the picture of the person face. The program should have a way to export a report of this database based on a giving Date & Time frame (FROM (Date Time) – TO (Date Time)). When the program is in drawing mode it should be in full screen for the current channel video. This program it should have the ability to have 4 and 8 Channels that should be identify in the source code e.g. marking a variable MAX_VIDEO = 4; or MAX_VIDEO = 8; , The program should have GPU ability for best performance: Our development environment is Windows 10-Pro, Visual Studio 2017, OpenCV 3.4.2 or greater, nVidia GeForce GTX 1060. The utilization of the OpenCV functions must be optimal, the language must be C++ or c#. Reporting should be like this: Camera ID, Person #, Direction, Date Time of Entry, Date Time of Exit, Date Time of Entry After DON’T WALK, Date Time of Exit After DON’T WALK. The last two data times markers will be only if pedestrian still crossing after the DON’T WALL signal is on, otherwise those value will be null. Imaginary report: 441USPARK01, 56221, UP, 12/7/2019 15:03:58, 12/7/2019 15:04:16,,[login to view URL] 441USPARK01, 56222, DN, 12/7/2019 15:03:50, 12/7/2019 15:04:10, 12/7/2019 15:03:55, 12/7/2019 15:04:10, [login to view URL] < THIS ONE IS a SAMPLE of person crossing after the DON’T WALK signal is on All source code must be transfer to our development environment and tested before We release any payment. We will have this bid to a single price in according to an agree. NDA and IP must be signed. We don’t have test videos We don’t have training data for Person Object.