The client requested the solution that will help to check and monitor the content displayed on outdoor ads (billboards). The main goal is to control the quality / legality of the displayed content.
We developed the solution, which is designed to check and monitor the content displayed on the moving billboard ads by using IP traffic cameras.
The solution consists of three parts - a common user interface (web page), a subsystem for video stream recognition (client), a subsystem for collecting and analyzing data (server).
The main content of the client page - the addresses of the videos and the column with current status (which displays the status of the video stream - whether it meets the standards / display no video X /, etc.) and direct links. The client can choose an address, video streams, and images with the ad content.
The general settings of the project include the ability to set the frequency of checking video clips for compliance, as well as the ability to add a new video stream. The server receives metadata, alerts about it, and displays the user interface.
The designed technology helps the Company to identify the most profitable locations with billboards and make pricing decisions. Technology stack: OpenCV, Tensorflow, Numpy, Flask, fine-tuned YOLOv3 object detection model, imagezmq library.