L&T Technology Services’ Video Analytics Solutions Help Manufacturers to Accurately Monitor the Field and its Assets

Video surveillance has come a long way from being a simple visual record of incidents to analysis of video streams in real time using complex algorithms, accurate object and event detection, and alarm generation. These applications are not limited to infrastructure surveillance of buildings and cities, and are even applicable to industrial units for asset security and management. In the security and surveillance area, L&T TS has developed Video analytic solution applications for event monitoring, object detection and object event interaction, video content analyzer, and human detection and tracking.

Industry Requirements: In industries, such as oil & gas, wherein the assets are widely distributed and not easily accessible, event monitoring, risk mitigation, and deviation detection are essential components of the security system to alert any suspicious activities in and around the field. Apart from oil & gas, these surveillance features find wide usage in overall industrial safety & security, automotive, transport, home automation, healthcare, and so forth.

An interesting and upcoming advancement in industrial safety & surveillance is video analytics, which can analyse both historical and real-time video streams. By analysing historical recorded situations, it helps in forensic analysis to identify events, attributes or patterns of behaviour. Moreover, it monitors video streams in real time and automatically generates alerts to prevent accidents or untoward incidents.

Background and Challenge: A major challenge for suppliers is developing intelligent cameras at a reasonable cost. Intelligent sensors and IoT-enabled advanced cameras are needed to get the most out of advanced content analysis softwares. These days, digital high-definition (HD) internet protocol surveillance cameras are replacing analog cameras because of lower installation costs, scalability, and the ability to add intelligence. In depends on the suppliers’ solution and its functional capabilities including accuracy and coverage, as to what extent video analytics can be utilized for industrial assets including oil & gas; and its interoperability with platforms such as plant asset management and building information modeling (BIM).

The Solution:
For industrial applications, L&T TS developed a video analytics prototype for capturing specific objects and events, and their interactions. The following step-wise approach was adopted:

• The video frames are processed frame by frame

• The motion features and object features are extracted

• These features are then sent to the classifier which recognizes the object and the event

   

   

After this, the user is alerted if there are any suspicious or outlier incidents. The classifier was simulated with various types of data, to make the algorithm more robust, accurate, and to reduce false alarms. L&T TS further optimized and tested the algorithm, and its performance was benchmarked.

Furthermore, computer automated image analysis or video surveillance automatically detects unusual events and trigger alarms. This reduces the volume of data presented to security personnel, and saves significant time and effort in taking actions.

Video surveillance solutions from L&T TS monitors large areas, handles complex query processing and time-varying images. Increasing the probability of detecting future incidents is equally important, and thus the specific video feed is shared with the surveillance personnel for post-event analysis for improved vigilance. L&T TS developed several surveillance applications such as video content analyzer, hand gesture recognition, human detection & tracking, and moving object detection as part of the video analytics solutions.

Moreover, the below mentioned suite of tools were used to develop these applications:

• Matlab, C++, Qt, Open CV

• Techniques like Histogram of Optical Flow; image processing algorithms and object detection features such as SIFT, SURF, HAAR like Features; Histogram of Gradients; Support Vector Machines; Cascade Classifiers, and so forth.

Quantifiable Benefits: The video analytics software and surveillance applications developed by L&T TS enable fast and accurate classification of objects and events. The algorithms are consistently refined by testing them with various types of data to reduce false alarms. In terms of return on investment and effectiveness of operational costs, advanced video analytics can provide a more reliable, functional, and automated video analysis as compared to simple motion detection.

Keywords: Asset Management, Motion Detection, Safety, Surveillance, Oil & Gas, L&T Technology Services, ARC Advisory Group, Video Analytics.