DailyPost 1748
VIDEO ANALYTICS – TRANSFORMATIONAL CAPABILITY
The amount of video being consumed on a regular basis for different critical activities of our existence has transformed it as the main medium of professional productive interaction. It also transforms into a proactive social tool. Digital transformation has changed the way we live. Education and work are all being conducted on this medium. The CCTV cameras of yesteryear are making greater sense because life has become touch less and so it has become an ideal tool for the Police work; crime investigation, law and order or intelligence gathering. From manufacturing to malls the intelligence these cameras would provide can change our existence. Today it is broadly in a passive mode. The plug and play which we desire is not going to happen.
This missing link or passively available technological marvel is known as Video Analytics. This has become a necessity as the world today is using data from video differently, and it is also being accumulated at a mind boggling pace. *This quantum of video cannot be handled at the manual level, however much we try; the conventional Video Command Centers and the prying eyes which keep guard. Life so to say has become a video. The screen has come right in the middle of all human operations. Out and out automation and actionable intelligence culled out of videos, can be the only sensible road ahead. From a policing point of view all the evidence would also come from it. It’s the third eye; unseen, ever present and vigilant, it becomes an ultimate tool for enforcement. The more the analytics, the better would be the quality of enforcement.
”The capability of automatically analyzing video to detect and determine temporal and spatial events is video analytics.” It transforms the video in real time into intelligent data. While the video is being recorded, it automatically generates descriptions of the happenings of the video on the technical side, is known as meta data. These can used for lots of intelligent purposes in conjunction with the video and even independent of it. Currently, CCTV networks are being used for variety of tasks namely; detect camera sabotage, real time crowd monitoring, movement tracking, counting & tracking objects, loitering detection, unattended / abandoned detection, missing object detection, video enhancement, missing object detection, automatic number plate detection to name a few. AI enabled video analytics failed to deliver in the Christchurch case and also faces immense challenges in case of doctored videos. DeepFakes seem to be insurmountable as of now.
Video Analytics is videos to actionable intelligence. Computer Vision with Deep Learning technologies make videos searchable, actionable and quantifiable. It can wipe out dead space out of videos and effective filters can lead to granular refinement. The number of filters have kept on increasing with the commercial use of Convoluted Neural Network Deep Learning over the last one decade. Computer vision has thus come of age. Direct benefits are right there. The error rate has come down considerably. False positives are on the way out.* Human behavior analytics is gaining ground. AI powered smoke and power detection can save life and forests. When rules fail which they will, CNN is the way forward. Convolutional Neural Networks are there to stay and will remain the mainstay of Video Analytics as we foresee. CNN generated Visual Similarity can sort out lots of identification problems.
VIDEO ANALYTICS IS THE REAL EYE OF THE VIDEO.
Sanjay Sahay