Low Resolution Long Distance Capabilities require LESS Camera Infrastructure

All iOmniscient algorithms are designed to operate with very low resolution video. While others tend to require a face to be presented with 60 to 120 pixels between the eyes for good recognition, the iOmniscient system can provide high accuracy with only 22 pixels and reasonable accuracy with as little as 12 pixels.

For behaviour analytics the system can operate with a video stream providing images with only 0.025 megapixels running at only 6 frames per second.

But don’t humans find it easier to watch videos with higher resolution? In an Autonomous System one should not expect humans to watch anything. The system analyses whatever needs to be analysed and can send a high resolution clip to the human for information.

No Expensive GPU Hardware

The use of much lower resolution cameras for everything means that one is processing fewer pixels, reducing the computing load. More importantly, iOmniscient has developed AI technology that goes beyond Deep Learning. We no longer need to label tens of thousands of data sets to train a system. We can do it with 10 to 50 images. The development of computing light algorithms eliminates the need for expensive, power hungry GPUs.

90% less storage and network bandwidth

iOmniscient introduced a Smart Compression for its video based systems that can store and transmit the unimportant parts of a video at low resolution while ensuring that the important information is always maintained at high resolution. The outcome is a 90% reduction in storage and bandwidth requirements for a 2 megapixel camera – more for higher resolution cameras. In many instances just the savings in storage is sufficient to pay for the cost of the intelligence required.

Fewer humans to operate the system

Studies have shown that it is ineffective to use large numbers of humans in a control room to watch computer monitors. With Autonomous systems finding the appropriate responder in the field, nearest to the incident and sending them information on where to go and what to do, it is possible to reduce the number of persons required in a large control centre by around 70 percent. Not only does this reduce costs but it provides the opportunity for more satisfying customer facing jobs for operators.