Gather more from video surveillance with analytics and deep learning

Gather more from video surveillance with analytics and deep learning
The ability to analyze and learn from security footages has become an important aspect of video surveillance. We examine several companies that developed video analytic software with features such as distinguishing different type of objects and learning capabilities to name a few.

Kipod Analytics Platform With Cloud, AI and Big Data

Kipod uses the power of the cloud, AI and big data to manage video content and footage analysis. With the cloud, the user is able to search through the footage of millions of cameras. Both the video and audio analytics is based on deep learning. The AI is able to identify specific persons and vehicles by tracking through multiple cameras. It also features alarm detection and object classification.
 

Searching past footages has been made more intuitive with the help of big data, allowing users to pinpoint specific content they are searching for. Kipod can search for footages with a specific type of vehicle, a person with a particular facial feature or even track audio that contain noises such as gun shot or glass breaking. Users can also set filters to only show video footages with certain attributes such as motion trajectory, obscured faces or stopping of certain objects.


Calipsa Deep Learning Video Analytics

Monitoring security surveillance for long periods of time can be an
exhausting task for humans. One can lose focus and concentration as time passes by. Calipsa developed an AI system that not only monitors and analyzes live feeds or recorded videos, but also constantly improves by learning from its mistakes. Human operators train the algorithm by rewarding it for correct notifications and penalizing for false alarms, ensuring that it never makes the same mistake twice.

If the operator is only interested in spotting a specific type of objects, they can train the AI to make the distinction. The system is sharp enough to distinguish between a pedestrian and cyclist, or varying sizes of vehicles on the road. It can also monitor and analyze multiple video footages simultaneously which increases the video coverage capability. All these functionalities are accessible by the operator from a web browser, eliminating the cost of dedicated hardware.

Sensority Suspect Detection Video Surveillance

Sensority has developed an analytic system based on cloud computing that incorporates the psycho-physiological aspect of people from connected surveillance cameras.
 
The Suspect Detection Video Surveillance (SDVS) processes hundreds of people in the video feed by combining video surveillance and psycho-physiological analysis to catch suspicious individuals or people. SDVS can make decisions and determine the subject’s activity through the pattern recognition method and by observing psycho-physiological changes over time. With deep learning, SDVS is able to improve the analytics over time.

SDVS can be connected to multiple cameras while tracking multiple subjects through each camera. The system will give a timeline of all suspicious activities and give an alarm list of suspicious subjects. It has a non-contact analysis distance of 10 meters, based solely on video stream. SDVS is ideal for areas that contain large crowd gatherings such as airports and stadiums, where large numbers of people need to be monitored to ensure safety and security.

Face++ Face Comparing

The technology behind Face++ Face Comparing is more than simple facial detection or recognition. It can compare two images for similarities by using deep learning, which helps with identity verification. The system is able to maintain high accuracy regardless of factors such as facial makeup, improper illumination and head position. Face++ helps to protect users from photo spoofing, fake faces and 3D avatar.

Utilizing machine learning, Face++ can analyze a person’s facial features to determine the age, gender, smile intensity, etc. To further increase the accuracy of the software, it can locate and return landmarks on a face such as face and nose contour. Lastly, the software can quickly search through photos and find similar faces, giving confidence scores and thresholds to evaluate the similarities. Face++ will greatly benefit settings where verification of documentation such as ID and passport is crucial.

Cogniac AI Visual Observation Platform for Inspection and Manufacturing

Cogniac AI Visual Observation uses an AI function that detects safety and security from visual feedbacks it has learned from user’s selections. By connecting the image and video streams, users can select the objects and conditions of their interest for the AI to focus on and generate automated observations within minutes.

The AI can be deployed in various settings to increase safety and have better security. It can spot workplace safety such as construction sites where worker needs to wear proper safety attire. The AI even monitors critical equipment and sends notifications if it sees conditions that may need attention. Security can be increased by giving the AI the type of image to send real-time alerts if an unauthorized action is detected. Cogniac AI can be used for much more than safety and security circumstances, users can also use the AI for logo recognition, terrain inspection or people counting.


Product Adopted:
Software
Share to:
Comments ( 0 )

asmag.com provides weekly and monthly e-Newsletters which include the latest security industry news, vertical solution case studies and product information.



Please key in code