The Importance of Simplicity for Video Analytics: Exploring ioimage

Published Jul 01, 2008 00:00 AM
PUBLIC - This article does not require an IPVM subscription. Feel free to share.

Maximizing simplicity will be a critical factor in IP Video success. In the early days, having a feature was enough. Now, with so many providers and IP trying to go mainstream, the importance of achieving simplicity will be paramount. I will use ioimage as a case study to examine how simplicity can be successfully employed.

[UPDATE: DVTel acquired ioimage in 2010.]

Simplicity Vital to Technology

The drive for simplicity is part of a much broader trend in all of technology. Technological success demands simplicity. The economics demands this. Complex solutions require human attention and labor. People are incredibly expensive. Even outsourcing is far more expensive than a technological solution that is truly automated.

Take Google for example. Google offers an incredibly simple user interface and incredibly simple way to find your result. Every request is automatically handled by networks of computers. Yet underneath the simple exterior is incredibly complexity. Nevertheless, the underlying complexity eliminates the need for human response to search requests while allowing you to get results to an incredibly varied range of requests. This is the paradox of complex simplicity at work.

The same goes for Apple products. The iPhone and the iPod are incredibly easy to use and yet they demand amazing engineering complexity to deliver such results. Yet the results are quite similar to Google – great satisfaction, minimal user effort and little human involvement.

The Challenge for Video Surveillance

Security managers have long accepted the potential of video analytics. The challenges have turned to more tactical and operational issues such as:

  • How well does it work? 
  • How hard is it to make work?
  • How hard is it to obtain?
  • How hard is to make work with what I have?

Just like 'anyone' can build an MP3 player, the value in delivering in video analytics is not so much in detecting objects but in solving the tactical and operational issues around delivering such a solution.

Let's examine how IoImage executed on each objective:

How Well Does it Work?

A number of strong indicators point to ioimage working well in the field. Among industry insiders, ioimage is very frequently cited as a company who stuff 'works.' This has been my experience both in my close personal circle and with broader industry acquaintances. Further testament to it is the dozen plus customer press releases [link no longer available] ioimage has issued. My experience is that customer press releases are very strong signs about the quality of a new product or technology. Not only does it indicate you have a customer, the customer is usually happy or they would be reluctant to go public with a release. While not having customer press releases does not indict a company, having customer press releases is definitely a bullish sign.

Nevertheless, what makes a product work well is more important than press releases or talk. Let's examine key design decisions that foster simplicity and higher performance in ioimage.

ioimage optimizes their solution for a single platform. While many video analytic systems offer their software on multiple vendor's PCs, DVRs, cameras and encoders, ioimage only provides analytics on its own cameras and encoders. Of course this limits customer flexibility. However, such constraints often make products simpler, more usable and more optimized to the task at hand. Indeed, a popular adage in the software development community is, ““Constraints are liberating.” Let us examine two ways this approach can make video analytics work better.

  • Focus Allows Maximization: Because IoImage only runs on one platform, a DSP, the entire solution can be optimized for it. By contrast, many video analytic systems are designed to run on one platform (e.g., ObjectVideo was first designed for PCs) and then transferred to run on another (e.g., TI's Da Vinci). While transferring to different platforms (i.e., porting) provides flexibility in customer platform selection, implementation issues between different platforms can significantly limit performance. Because ioimage started and focuses solely on a DSP solution, it can maximize the solution for that platform.
  • Control Maximizes Performance: A key factor in the success of an iPod is that Apple designed the entire solution. It controls all aspects of the hardware and software. Apple does not need to worry about differences in hardware specifications or setup on different devices. It can stringently handle and confidently solve problems across the entire product. In this sense, the IoImage is like an iPod because ioimage can control and ensure that it solves problems end to end. By contrast, with the popular alternative model (loading software on third party devices), the video analytic vendor has serious constraints on optimizations and could suffer from problems in the other components.

How Hard is it to Make Work?

ioimage provides a graphical point and click user interface that is quite similar to the graphic applications that millions of people use everyday. What's interesting about this is that this is not common today in video analytics. Most video analytic configuration tools resemble the controls of a jumbo jet. For the next demo you take, go ahead and ask to see the configuration tools. I am no longer surprised about seeing systems that offer over 200 configuration parameters with cryptic names and impossible to guess relationships. The ease of setup is a significant differentiator.

Nevertheless, making a simple user interface is actually incredibly hard work. Companies do not build hard to use interfaces because they do not care. Usually such problematic interfaces are a product of a lack of resources. Building a good UI can take 3 to 5 times the amount of design and resources as a bad UI. This is a core example of the complexity needed to deliver simplicity.

Having used ioimage's UI in the field, you do need to understand the high level concepts involved just like you would need to understand the basics of drawing to use a graphics program. However, unlike the jumbo jet type controls in many video analytics, you are freed from understanding lower level issues that someone in the field usually will not know.

How Hard is It to Obtain?

ioimage's product design and distribution has made obtaining the solution simple.

The product has been designed to combine the video source and the analytics into a single appliance. As such, this eliminates a costly and risky element of bringing those two elements together.

Distribution is incredibly easy as literally anyone can obtain the products. Not only can thousands of security integrators purchase the product from Northern Video, installers and integrators can buy directly from Super Circuits.

Contrast this to the Byzantium product purchasing process of Cisco or the politicized negotiating of dealerships for legacy security manufacturers. ioimage has made its products available essentially ubiquitously.

How Hard is it to Make it Work with What I have?

System integration has been one element where ioimage has historically not been simple to use. In the past, ioimage's API required detailed programmatic knowledge that only advanced programmers possessed. Even with such programmers, integrations were time consuming. This limited the ease and number of partners integrated with ioimage.

Last year, to correct this, ioimage introduced a “web-based” API. Millions of programmers today are integrating different web applications like Google, Amazon, Ebay, Facebook, Twitter. They all use similar protocols and processes to make integrations super quick and easy. ioimage has leveraged this same approach in its API. This now allows a vast community of programmers to perform ioimage integrations. It also makes integrations quicker and less painful. (Note: ObjectVideo offers a similar style API in its OVReady offering.)

Much like a 'simple' user interface, a simple API is actually harder to implement because the programmer must properly hide details while ensuring integrations can work smoothly.

Furthermore, ioimage has asserted that they will not politically restrict who has access to their APIs. While you need to sign an NDA to use, anyone can do so. This is a big benefit as it eliminates the biggest historical problem of video surveillance integration – using APIs as a sales weapon.