Sunday 8 December 2013

Risks and challenges of video analytics.


Video Analytics as a whole is a relatively new technique, therefore the impact it will have in the future is still relatively uncertain.

Let´s look at this as if it were an emerging country: A lot of opportunities to take advantages from but also some risks that should be taken into account.

During these months we tried to explain the main topics related to video analytics all the while being aware that, some applications could carry some precautions.

Nowadays we tend to trust our devices but… would you trust to 100% a car that is being driven by itself? Would you feel safe taking curves on a rainy day?

Not every alarm received from our fall detectors will mean a suicide and not every suspicious behavior that our camera is detecting will be a terrorist trying to place a bomb. Technology works for us, but it is our duty to filter and control this information.

At the same time that technology is evolving, the applications we make of it are increasing too. So it is easy to guess that in 10 years a lot of different branches will have grown from the core idea of Video Analytics.

The main challenge for the future is not to come up with new inventions, if not making them completely reliable.

For example, in terms of security we live in a world where new threats are arising every day: imagine that someone can potentially hack your Google Car and drives it out of your sight.

From SmarterView, we cannot predict the future inventions related to technology, but two things are clear: Video Analytics trends will squeeze them to improve our welfare and, above all, SmarterView will be here to tell you the news.


Motion Capture Analysis

  Video Motion is a branch of Video Analytics that intends to create a virtual image based on real motions.

            Where can we implement these techniques?

  Since decades ago the film industry has been connecting animations and virtual scenes with real motions. Gone are the times of drawing every scene with a pencil. By taking advantage of these technics the quality of films has been improved. As an example, everyone can remember “The Lord of the rings” where one of its more popular characters, Gollum, was completely made of Video Motions’ technics.

 Also some companies related to video games are using Video Motion. Through transfer the real movements from the actor that is embodying a super hero or from the NBA player, we make the game experience much more realistic and valuable for the consumers. We now introduce one leading company in motion capturing industry.




Established in 1982, Motion analysis is leading company in the industry for 3D passive optical motion capture. It develops and provides motion capture hardware and software around world. Such motion capture equipment is using in various sector, such as entertainment, movement analysis, and industrials.

  One of Motion analysis’s remarkable products is eagle digital system, the first system applied digital camera method, and it opened the era of digitalization in motion capture industry. This system provides real time value so that more accurate high quality data could be delivered to analyst with reducing cost and time.



  Their works in entertainment industry includes Film, video games, broadcast, and animation.  Using MAC system, producer could capture multiple actors with their body movements, while the system dramatically reduce the cost and time.  Movie Matrix, Iron man 2, and Final fantasy were created by MAC system.





  In the movement analysis sector, Motion analysis specializes in gait analysis, rehabilitation, sport performance, and medical robotics. With Motion Analysis Real time system, it provides state-of-the-art motion capture, which allow the analyst to quantify the interval of extremity function in Patients with stroke, head injury, and Parkinson’s diseases. 




   source: 1) http://www.motionanalysis.com/html/movement/movement.html
                2)http://www.youtube.com/watch?v=DVQ9MPjk5dI
                3)http://en.wikipedia.org/wiki/Motion_capture
                4)http://en.wikipedia.org/wiki/Video_motion_analysis







Tuesday 26 November 2013

Exclusive Interview with Dr. Alexandre Alahi Member of the Stanford Visio Lab and Founder of VisioSafe



Smarterview Exclusive Interview with Dr. Alexandre Alahi


The SmartView team is proud to release its first interview of Dr Alexandre Alahi expert in video analytics and member of the Stanford Visio Lab. Alexandre is also the founder of VisioSafe, the first hosted video protection service with avant-garde analytics. He has also written several research papers and articles. Some of them have been published in world famous newspapers such as the WallStreet Journal. Last but not least he won several awards in different contests such as the Swissnex Boston Global Pitchfest or the ICDS challenge prize. 
Hi Alexandre how are you doing? Thank you for your time and consideration. Let us start off right away. You are one of the founders of VisioSafe. Could you briefly describe what your startup is about?
The goal of our startup is to add another dimension to Google Analytics. Google Analytics is used to analyze the traffic of visitors on your website and what we want to do with VisioSafe, is to analyze visitors in the physical world. We want to fill the gap between the web and the real world. The success of online e businesses is thanks to analytics. They make you understand how people navigate on your website, in other words; what’s their path, how long they stay and what the impact might be of a given layout. We are applying the same tools for the physical world so a company can make the same analysis and optimizations.
Due to that we transform cameras in analytic tools and measure the traffic of people in any area. We use artificial intelligence to automatically detect and track them on the ground and analyze their trajectories in real time. The main advantage is that we re not embedding any devices on the visitors and its also privacy safe since we only capture their position and coordinates. We don’t track any info about their identities.
Who are your potential clients and how do they benefit from your technology?
Our clients are retailers and any brick and mortar companies, malls, exhibitions, airports etc. By the way, we call our cameras sensors as we use them for data collection. For retailers the benefit is pretty clear as why to use our technology. Understanding the purchasing path, the fishing rate, how many people come into your store and do not buy something, identifying hot spots in your store and how to optimize your layout. This data is most valuable for retailers and it is the same for malls and exhibitions. Which stores are generating the most traffic in order to optimize the placement of your products or stores depending on the traffic.
Airports have the same goal as train terminals have. They want to reduce the congestion time so you don’t miss the transit. They generally invest a lot of money in rebuilding their infrastructure. But for them to do so, they need prior quantitative analysis to justify the corrections. We are diagnosing the situation and also identifying how to improve it. Changing infrastructure in airports or train stations is really expensive so they have to do it right. However in retail stores the cost is much less.  Still the use of video analytics must be cleverly conducted. If it is, video analytics is the tool that can help to add value.
Can you show us an example where your company made a significant impact?
So this information is confidential. But to give a broad idea, some clients have doubled their revenues by only changing the layout thanks to the data collected. It is generally known that the way you present your products has a huge impact on the purchasing process and revenue. Some stores really make you feel uncomfortable. You enter and you want to leave again immediately. Others however manage keep customers in their stores, make you find the products you look for easily and maybe even make you buy things which you were not even specifically looking for. Hence optimizing the layout is definitely very important in terms of revenue. On top of that, the optimization of staff allocation is pretty important too. How much staff do you need when and where? Here you can surely save a lot of money. Moreover, analyzing queues and waiting times is critical for airports and train stations. Simple experiments, e.g. changing light conditions, can help to measure the impact on the flow of people.
The good thing is that our company offers a fairly new product and or service which people are not yet used to. So the clear value we are adding, pretty quickly convinces customers to follow up with repeated purchases.
What are the key management challenges when setting up/implementing your technology?
We are offering a cutting edge technology, which is a result of several years of research. It’s a new service thanks to the break troughs in artificial intelligence. People are actually still baffled and impressed when we demonstrate the capabilities of our product. Let me give you an example with regards to train stations. When we met with the train station officials for the first time, they tried to find other companies in the world who could do the same at the same scale. At that time we were the only ones. Again, our advantage is that we can locate as close as to a few centimeters. As opposed to a few meters when locating cell phones. And we are doing it large scale. We are able to track people over a minute or over one hundred meters. These are pretty nice features I would say, which others are not offering.
Let’s talk about personal data. Tesco is for example planning to scan faces of customers to target advertising in store. How and where do you draw the line between observation and infringing personal rights?
We all use GPS or elevators… Again, we are just offering a tool that helps potential customers to analyze the occupancy of their sites. And all that we do without identifying people. We don’t know whether it is a woman or a man, we don’t know their age… We only know about people and their trajectories. And we don’t want to know on purpose. We don’t want information about people’s identities. Of course when it comes to mobiles for instance you somehow get identities. And whether that’s good or bad is another debate. But our technology is not intrusive. Anyway, some of our customers have cameras installed anyways, for security reasons. And they are capturing images. We also offer products that help with that respect. But the analytic tool does not need images. It is the choice of the client whether or not to capture image data for security reasons. Some already have cameras installed and we are just helping them to transform their devices into an analytical tool.
Last but not least what do you think are the future trends regarding video analytics?
This is a topic, which we have been working on researching for he past 20 years. Every year the cost of the cameras decrease. Already two years ago there were 50 million security cameras around the globe.  All this follows the major topic of big data. The challenge is how to process it! Right? We have all the data. But we need artificial intelligence, a software or machine, which is able to understand and act. This pretty much describes my professional field and that’s what we are trying to do: Creating smart cameras. And there is definitely still room to make the cameras even smarter, to make them more human. We are still very far away from what a human brain can do, but at least we are trying to fill this gap. Cameras need to be able to cluster, recognize, understand and retrieve information from data. Because in the end it is only numbers, right? You want to obtain semantic information. You don’t want to receive an alert that someone is walking; you want to receive an alert about abnormal behavior.
Well thank you very much Alexandre. That was pretty insightful.
For all of you that want to stay updated what Alexandre and VisioSafe are up to in the future, be sure to check out these links:

Saturday 23 November 2013



Following decades of slow market penetration and setbacks, the Intelligent Video Surveillance, ISR Analytics and Video Analytics industry is forecasted to experience decades of rapid growth. In the past year, intelligent video surveillance, ISR and video analytics with mature technology have attracted a lot of attention. The technology, which automatically identifies people and objects in video content, has matured to the point of practicality and is now winning business adoption, according to another report from technology research firm ABI.
The Intelligent Video Surveillance, ISR & Video Analytics: Technologies & Global Market – 2013-2020 report indicates that the global Intelligent Video Surveillance (IVS), Intelligence, Surveillance & Reconnaissance (ISR) and Video Analytics (VA) industry revenues* totalled $13.5 billion in 2012, and are expected to grow at 13.8% CAGR from 2012 to 2020.



Features like motion detection and camera tamper, embedded in surveillance devices and offered as free by manufacturers, will be increasingly joined by applications for more sophisticated video content analysis (VCA). There are also technology improvements such as (fewer) false alarms, searchable analytics and increasing processing power.
Examples of applications of video analytics include recognising human faces in speeding cars, people counting, identifying customer traffic patterns, assessing customer dwell times at different areas of a shop, analysing the length of queue lines etc.
The rapid market growth is driven by:Technology maturity, Increased use of video surveillance due to high processing rate of video surveillance compared to human operators who entail high cost and have a high rate of overlooked events, Migration from analog to digital and IP-based cameras, Cost reduction of video analytic systems and Improved cost-performance of new edge-based video analytics DSP technologies.


Sources:

Saturday 16 November 2013

Video analytics in retail another point of view.




Not that long ago, it wasn’t uncommon for retail outlets to have surveillance cameras stacked up in their corners to monitor customer movement with the primary purpose of preventing theft and catching shop lifters. Nowadays, retailers have discovered that this practice of monitoring customer movement, known as video analytics can also be used to serve as business process monitors, providing valuable data and insights that offer real benefit to the operations and people they serve.

Today, video analytics systems have become more sophisticated and technologically advanced and can significantly empower retailers to maximize efficiency whilst monitoring customer traffic and in-store customer behaviour, improving operational efficiency, minimizing theft and loss, increasing customer safety and ultimately improving sales and increasing profit.


By expanding the profile of the cameras, consumer researchers can analyse and push out dashboard reports on consumer behaviour. Businesses can better understand market behaviour in stores, what drives a person to a specific spot in the store, who is busy in the store at specific times of the day, age group trends, what is the consumer buying, the time they spent in a specific area of the shop etc.


How does it work?

With advanced motion detection and pattern recognition, Video analytics software examines each pixel in the frame and picks up even the slightest movement. Specific patterns can also be programmed for recognition.





With Video Analytics Retailers can:

  • Use heat maps from advanced video equipment to ascertain the areas with the greatest frequency and activity (“hot zones”)
  •  Analyze store traffic to identify the dominant traffic paths
  • Assess the effectiveness of window and marketing displays by measuring customer dwell time
  • Use queue data to enhance operational efficiency and increase customer satisfaction
  • Use surveillance networks to prevent accidents and enhance safety measures, detect suspicious in-store activity and monitor suspicious after-hours activity

Benefits
  • Helps to pinpoint premium product positioning, determine general strategies for product placement and make better merchandising decisions
  • Informs marketing decisions such as the most appropriate locations for in-store promotional campaigns
  • Indicates the success of product placement and ROI on advertising investment
  • Makes decisions that optimize staff allocation, space utilization and traffic flow in line with data regarding dwell time (stickiness) in certain areas 
  • Understands the customer traffic at check-out points and product counters across different time periods and seasons, to more strategically allocate human resources across the store and optimize staff shifts
  • Detects suspicious activities from customers and even employees


The ability to accurately see where and what people are doing throughout the mall gives management a greater capacity to cater for customer and tenant needs

Sources:

Wednesday 13 November 2013

Can Video Analytics stop suicide?

Every year more than 1000 people around the world die inside of Metro railway stations.  The use of Video Analytics can help to identify and prevent suicide attempts. You might wonder how? The answer is very simple: falling detectors.

Falling detectors are used in public and private places for automatic detection of abnormal situations related to falling objects. It is important for high-quality security and reduces the numbers of severe accidents.

Tasks solved by falling detectors may be divided into two categories:
The detection of everything that falls from the platform and the detection of unintentional falls whilst on the platform (serious injuries and citizens living alone). In both cases immediate treatment is very critical, thus falls need to be detected as soon as possible.

In these two cases detectors work in different ways.

In the first case the detection logic is based on signal line intersection. This line is the boundary of the platform and the rails whereby the intersection of it is secured by an alarm signal. Hence the fall detector can automatically track all abnormal situations associated with falling of objects. It increases the speed of response and helps to decrease the number of death.




In the second case the evaluation of falls is more difficult as there is no dangerous boundaries and body movements and the fall is unpredictable. The principle of the fall detection in this case is based on a combination of motion gradients and human shaped features variation. Large movements always accompany falls, but large motion can also be a characteristic of walking person. That is why we need further analysis. An analysis of changes in the human shape. Especially width to height ratio up to a certain threshold is considered to detect a fall. And the last analysis is a luck of motion just a few seconds after the fall. All these steps help to identify falls and can trigger alarm signals in order to provide medical help quickly.



What are the main benefits of fall detection?

Video analytics helps timely response to the fact of people fall in the observed area and helps to providing quick treatment, which in most cases is critical. Therefore it helps to decrease the number of accidents and may save lives.




References:





Sunday 3 November 2013

Handsfree driving

The future of driving with the Google Self Driving Car. How video analytics will make your life easier.

Planning your next road trip? Imagine a world where you could just enter in your car, set up a destination, and then relax. This is almost real!


The first to imagine such car was Norman bel Geddes who designed General Motors’ Futurama exhibit for the ‘39 World’s Fair in New York’. Consider as a dream as this time, it is on the way of becoming real. Google’s project, which is led by Sebastian Thrun, former director of Stanford’s Artificial Intelligence Lab, is one of the most promising improvements for the car industry.
As Larry Burns, former GM research chief, professor at University of Michigan and a Google adviser said:  “Autonomy changes everything. Giving automobiles the ability to drive themselves is the biggest thing to happen to the automobile since the automobile".

How does it work?

Google self-driving car is an extension of the already existing driver aids that you can find in some luxury cars such as the BMW I3 or Mercedes.  The techniques used, are lane keeping systems, adaptive cruise control system, auto parking system, emergency braking and satellite navigation systems.  The driverless car is basically using all these systems together with the help of specifics software. The car covered with sensors and cameras is also able to keep track of the surrounding environment. Every element such as lane marking, reading signs, traffic lights and pedestrians are detected and analyzed with the help of cameras, radars, lidars (specific type of radar) and video analytics software.
Moreover in order to have an accurate positioning the Google self-driving car is using specific technology such as gyroscopes and accelerometers. It also scans its surroundings and collects as much data as possible, which are then analyzed in ordered to extract precious data in case of accidents.



What are the benefits of such innovations?

Such innovations will completely change the relationship we currently have with the automobile. It will also reduce drastically the number of car accidents, reduce the traffic flow and allow disable or old people to finally use their car. Also application such as Uber would be able to increase its efficiency and reduce its costs by spreading their driverless car fleet in important cities around the world.
What is the main issue?
However there is some concern related to this major innovation. The main challenge for autonomous vehicles is the way they communicate and interact with other cars. Several situations such as deciding who should proceed first around an obstruction needs a great deal of situation analysis and elaborated video analytics software.
To conclude video analytics is again used for a major innovation that will probably change our everyday life. Combined with other technology video analytics can really improved our daily life and object that are today consider common will maybe be perceived in a complete different way tomorrow.

References: