Intelligent cameras provide actionable smart data to improve safety, increase efficiency and reduce business costs. So why is it that not many businesses use video analytics?

Many end users are not aware of what exactly video analytics is or does and hence do not ask for it. They may also not fully understand what analytics can do for them. On the installer side, video analytics takes quite some time to set up properly, so there’s the time to get educated in the first place, and then the extra time for each install. Jobs can take much longer if the analytic requirements and the environment are challenging.

There is also a perception that video analytics is not reliable enough, or it will put video surveillance monitoring professionals out of a job. Here, the statement is fractionally true. Some video analytics are very basic and can only be used for simple detection such as motion in a region (most likely in a sweet spot close to the camera). And majority will only work well indoors where there are no environmental “distractions” like rain, tree movement, light changes. Advanced video analytics engines are much more complex and reliable allowing immediate detection and classification of objects across the majority of their field of view, no matter the environmental conditions. On the human replacement side, it is our view that video analytics only improves what is being done already, rather than displaces security professionals. Small businesses do not employ security professionals, hence analytics can provide alerts to the end user via smart devices for example, where there would be none in the first place. For organisations where 24-hour video monitoring is essential, video analytics helps the operators to see and action events of interest quicker and more reliably, and also find recorded events of interest at drastically faster speeds.

The last objection we hear about video analytics is the cost, and yes, the cost can vary greatly. There are what some call analytics in cameras (on the edge analytics) costing as little as $100. We will not comment on their analytic robustness. There are also server-based analytics that generally come with substantial licensing costs, analytics modules and possible software customisation. This is the expensive side of the spectrum, only surpassed by totally custom written software. There is a middle ground however, where the video analytics are quite reliable and do not cost the earth.

So, to the first statement, how can video analytics improve safety, increase efficiency, and reduce business costs. Here are several examples.

Typical Security Surveillance Analytics

• Perimeter Protection – Get alerts when objects are identified near or crossing a perimeter. Distance depends on the camera but can be up to hundreds of metres.
• Forensic Search – For example, show all instances of yellow cars turning left within a specified period of time.
• Camera Tamper/Vandalism – Send alerts when cameras are being sprayed, blocked or vandalised.
• Crowd Detection – alerts for build up of people in an area
• Vibration Reduction – ideal for cameras mounted on poles, bridges, heavy machinery
• Object tracking – track a person or a moving object (within the field of view or from camera to camera)

• People Counting and Occupancy – Count people entering and leaving to give net occupancy
• Regions of Interest – For example sound alerts when an individual in a factory or warehouse goes outside of permitted walking path
• Door Obstruction – Identification of objects blocking doors or isles or pathways.
• Object Removal – Alerts when objects are moved from their position
• Queue Lengths – Alerts when a queue at a counter gets to a certain length
• Vehicle Entry, Exit & Counting – Ideal for logistics and manufacturing – detect trucks at gate and count entries and exits

• Traffic Count – directional identification and count of pedestrians, bicycles, motorbikes, cars, trucks and buses.
• Illegal Parking Alerts – Vehicles stopping in no-stop designated areas
• Wrong Way Alerts – Vehicles driving in the wrong direction
• Idle Objects – Alerts for objects left alone in a region of interest for more than a specified time (at airports and other travel hubs for example) or debris alerts on roads, runways
• Identification and Tracking of Craft on Water – one of the hardest video analytics tasks due to changing nature of water and waves.

Apart from the quality, data security and performance, all current Bosch video cameras from the 3000i range up feature video analytics in the price of the camera. From 3000i through to the 6000 ranges all Bosch cameras features Essential Video Analytics. This set of analytics is ideal for indoor and some controlled, short-distance outdoor applications. IVA or Intelligent Video Analytics are built into the 7000, 8000 and 9000 camera ranges and are ideal for indoor or outdoor applications with longer distances. The Intelligent Video Analytics are very robust and take environmental conditions into account. The forensic search engine, based on metadata information, is one of the best in the business, finding objects and events of interest in seconds. The two major advantages of using analytics on the edge (in the camera rather than on the server) are reduced cost of acquisition and reduced server computational and storage power.

Camera Trainer is a free machine-learning Video Analytics extension available for Bosch IVA cameras. It allows you to teach the camera to recognise objects and situations A simple example is learning to recognise an empty car park spot to an occupied one and based on this providing a count of occupied and empty car spaces.