Companies invested in power production owns massive size of power assets. They roll in and out billions of USD, thousands of employees and associates, through mammoth infrastructure, spread across many geographical locations. Besides aiming maximum returns to stakeholders, they are consciously invested in contributing to infrastructural goals of the country and their own corporate social responsibility. Technology enabled Intelligent management systems run like veins in the entire complex structure. Lot is at stake. Vulnerabilities and risks are as many.
Video Analytics is one of the many areas they look into to have a technology edge to smartly control potential risks and threats, lurking around the limitations and fallibility of security and surveillance systems installed. Video surveillance is a very old task in security domain and has carried its itches. However, much water has passed under the bridge, from the times of human-monitored exclusive systems to current video analytics based solutions, where a human is needed to monitor the alerts generated by a video analysis system and decide what should be done, if anything.
Besides perimeter security, their backyards, mine areas, warehouses, storages, normally have hundreds of CCTV cameras installed throughout. These surveillance cameras observe the many processes within the area, and are mostly installed in a permanent, overt manner, with portable covert cameras being used on a temporary basis. The control room operators observe the general movement of objects and people, but with limited success.
They have too many security cameras to monitor and cannot effectively highlight or detect unusual, high-risk or tell-tale events. This can betray suspicious activities, leading to thefts, huge loss of resources, and associated impact on related business activities.
However, the maturely trained intelligent video management systems can provide strategic assistance and integrate with virtually any source of event data. Complex rules can be constructed to take into account related, but disparate events, which indicate impending failure or situations that warrant further investigation. These enable the video operator’s attention to be targeted much more effectively to potentially productive areas.
We found that an intelligent Video Analytics software could contribute in a major way by providing means of accurately dealing with volumes of data, information and scenarios. And we provided that solution.
At VertexPlus, we developed a video analytics platform to be able to process high volumes of data, detect intrusions, monitor vehicle movement in real-time, and give relevant statistics to the client to be able to take actions in areas where suspicions were abundant generating theft issues.
Vehicle counting, differentiating between models of vehicles, developing anti-theft mechanisms, training system on specific scenarios, generate high-value statistics used to obtain insights about movements and activities that commit an infraction.
To reveal the potential for adopting and adapting to systems in use at a mining environment we first needed to look at systems already employed. And to have a clear view of the entire area from various angles, where the events being monitored might occur.
The scenarios to focus on, were defined. Such as processing Live Camera Stream from IP Camera and Archive Video Processing
The data came from various video streaming sources such as CCTV cameras, Industrial IP Camera, Cloud Video Stream, Archived Video Stream etc.
We used pre-trained model as well as custom model which could be trained
Video analysis software was run centrally on servers located in the monitoring station. This is known as central processing. Central Processing refers to where data is processed centrally or remotely.
The system must provide valuable support to human operators by helping them , detect events that might otherwise be overlooked or take a long time to detect manually.