meta data
Metadata is feature attribute information extracted from a target object which can be
used for data retrieval. Currently, there are three main kinds of metadata in the
security industry: human face, human body, and vehicle metadata. Facial information
includes sex, age, glasses, masks, expressions, beards, etc. Human body information includes tops, pants, clothing color, hair, backpacks, etc. Vehicle information includes license plate, color, brand, model, etc.
people counting
The number of people that enter/leave/pass some specific areas within specific time periods is counted after filtering out non-important targets (shopping carts, wandering personnel, etc.) Retail analytics application.
facial recognition
The face recognition function can be used to determine whether faces are present in the input face image or video. If faces are present, the position and size of each face and the position of the main features of each face are further given, and according to the information, the identity characteristics of each face are extracted and the human face will be modeled. Each face model is compared with face models stored in the known faces database to identify each face. There are three methods of comparison: 1:1/ 1:N/ N:N.
Deep Learning Applications
In the surveillance industry, primary target objects of deep learning algorithms are people and vehicles. Taking this into account, DTEQ provides the following technology suitable for various applications.
More Applications on Deep Learning
FALSE ALARM FILTER
Further analysis is performed on detected behaviors or events, automatically filtering out false alarms introduced by animals, rustling leaves, bright lights, rain or snow, etc., greatly improving alarm accuracy.
ANPR
ANPR (Automatic Number Plate Recognition) is a technology that uses optical character recognition on images to read license plates with high recognition accuracy.. ANPR applications include toll collection, traffic monitoring and security, speed and journey time measurement, parking and access control, etc
INTELLIGENT IMAGE SEARCH
.Image search function refers to the use of facial, human body, or vehicle images to search for related pictures and video information
FLOW PATH
Video frames from Deep Sense cameras at different locations are integrated to trace target movement via image matching and image searching techniques.
Deep Learning Features in AI
1. Deeper.
Compared with traditional algorithms, deep learning algorithms have a deeper
structural level. In addition to the common input and output layers, there are more
hidden layers in the middle, from the underlying features to more abstract
high-level attributes or feature extraction
2. Higher Accuracy
In this era of massive data and computations, deep learning follows a positive cycle
chain – the larger the data, the higher the accuracy of the algorithm, the higher the
accuracy rate, and the more accurate the data collected.
3. Higher Flexibility
Deep learning algorithms are enhanced by training and learning, and can be adjusted
quickly and adapt to various new problems. They can learn to identify more object types.
Big Data
Now you understand that AI is powered by machine learning and deep learning, and none of it can function without data. Imagine this scenario: you have millions of terabytes of data that are virtually useless unless someone can find a correlation between the data. This is exactly what happens in a video surveillance system that is not using machine learning or deep learning to pull meaningful information out of the data that’s being collected. Big data combined with deep learning, on the other hand, has the potential to transform video surveillance from a passive visual surveillance solution to a much more active one.
Here’s an example. Using a people-counting camera can provide valuable information on how many people are entering and exiting a defined area. It can also detect loitering, keep capacity counts, and signal an alarm when pre-defined thresholds are exceeded. Combine that with Point of Sale (POS) information, and counting how many people pass by certain retail displays can provide valuable insight. You could better determine the effectiveness of an end-cap display, such as tabulating how many people stop to look, verses those who don’t stop and compare it to actual purchases. Adding facial recognition can help detect additional patterns in browsing that would otherwise go unnoticed, such as men with beards stopping and looking at female-oriented product displays. Whether you understand why this is happening, it illuminates an interest trend nonetheless.
Now that you know the differences between AI, deep learning, and machine learning, and understand the role that big data plays in these technologies, you can better implement them in your video surveillance system.