Saffree: The Second Sprint

In the second sprint of the Saffree project, my team and I have further improved our Saffree system. The 1 meter distancing has been completed, a basic flask website is also created to view the amount of people in each venue, and the Saffree Box has a new feature to detect SafeEntry confirmation pages.

I have added a new functionality which allows the Saffree Box the scan for SafeEntry confirmation pages. The way this was done is that the Saffree Box will scan if anyone is in-front of the box using an untrasonic sensor. If a person was detected, the program will then make use of the camera on the box to take in a live video feed from infront of the box.

For each of the frames recorded by the camera, the program will remove all colors except for green. After that is completed, I made it so that contours will be recorded within the frame, so that the coordinates of the contours around the green parts will be saved. Using those coordinates, I can crop out the green areas from the original frame, and the cropped out areas will be put under an word detection algorithm, in order to check whether the object really is a SafeEntry confirmation page.

As seen in the example above, the first image will be the original frame taken in by the Saffree Box, and the program will then remove all the parts which are not green. From there the contours coordinates are saved (represented as the red outline in the image on the right), and text detection is applied on those areas wrapped by the red outlines.

By doing so, the Saffree Box is able to detect SafeEntry confirmation pages.

Once the Saffree Box detects the SafeEntry confirmation page, the system will record the visitor as "Checked-In", which will allow the visitor to enter the venue without sounding an alarm.

Other than this feature, I have also improved the Saffree Box software as a whole. I threaded the camera stream, so that the video input will always be in sync even if the processing speed is slower than the video stream. I have also enhanced the socket server, as well as the barcode scanner.

Sprint 2 Backlog:

As the scrum master, I also ensured that the scrum meetings are consistent, and kept track on each of my member's progress. After realizing that some of my members needed help, I assigned other members who have already completed their tasks to aid those in need.

However, the planning for this sprint was poorly done, and my team was overloaded with work to complete, resulting in certain tasks having to be pushed back to the next sprint. Hence, in our sprint retrospective for this sprint, we have discussed to set more realistic goals for the next sprint, to avoid having to rush the tasks.