Andy is one of the most respected forensic video analysts in the UK. In this interview, he shares his point of view on forensic video analysis and the industry at large. I want to highlight his very insightful observation about the situation we have been fighting since our beginning: “A widespread fundamental misunderstanding about the complexity of audio-visual forensics work.”
Enjoy the read!
Martino Jerian, Amped Software CEO and Founder
Andrew, tell us a bit about yourself. What is your background and your current role at HM Revenue & Customs?
I’ve been around a long time. My career started in 1983 with military communications engineering in the Royal Air Force where I did seven years, four of which were spent in Germany which I loved.
I then moved into broadcast television where I spent twelve years managing technical teams such as TV camera crews, sound and vision engineers, directors, production assistants, technicians and video editors for a number of companies including ITV, Sky TV and a number of outside broadcast companies. I used to design and build TV studios, galleries, transmission facilities and OB trucks.
Following redundancy from one job I saw a role advertised in the police, managing teams of audio-visual forensic specialists, so that opened up my current career where I’ve spent 16 years in West Yorkshire Police – and I’ve been working with HM Revenue & Customs for two years now.
What made you decide to enter the field of multimedia forensics?
In all honesty, it never occurred to me that such a career existed. Having spent time in the TV world I had no idea that the police used video (and audio, photography and CCTV) in a forensic way.
It was only when I switched career that I realized there was a whole new world of opportunities beyond TV, and I have genuinely never looked back. The big difference is the feeling that you’re doing something worthwhile, using technical skills in a way that makes a difference in society. That sense of civic duty is something I share with many colleagues, and this is why we love what we do.
What would you say are the biggest challenges with video evidence during investigations and when presenting it in court?
In the UK policing world, there are two major challenges. The first, sadly, is financial. Not having budgets to properly train and develop staff, invest in the latest technology, or provide sufficient resources to meet the demands are all very real and difficult problems for police managers.
The second is volumes of data. Video evidence is everywhere these days. From CCTV in public and private spaces, to doorbell cameras, dash-cams and helmet cameras on cyclists, to body-worn video on police and other officials, mobile phone videos from members of the public and social media content. And with image quality continuing to improve, HD and UHD video is resulting in huge demands on data storage and processing.
Since changing jobs, I have found that within HMRC, unlike many in the wider law-enforcement community, there is a genuine desire to invest in staff, provide training and development and provide the tools we need to do our job to a world-class standard.
What would you say are the main forensic challenges surrounding image validation? How can they be addressed?
Keeping up with technology is always a challenge. It always amazes me how every new technology can be very quickly picked up and utilized in criminal ways, often before it becomes mainstream technology.
In the current climate, social media imagery, for example, is so difficult to validate. With the growth in augmented reality and AI/machine learning capability this is becoming an ever more difficult challenge.
I think the real problems for law enforcement are yet to come and will only be addressed when new tools are developed, which leverage emerging technologies in a different and innovative way.
To achieve this, two things are required: real and significant investment in audio-visual forensics, and genuine recognition of the problems with a desire to solve them.
In your opinion, how important is it that digital forensic techniques and tools are based on the scientific method?
I think it’s essential. The scientific method not only provides rigor and quality assurance, but it also provides gravitas.
Future investment in crime fighting will come from the need to tackle digital crime: cyber, cyber-enabled, digital forensics, audio-visual forensics, mobile phone work, etc. It’s a growth area. This investment will only happen if the forensic community is seen to be operating in a professional environment and taken seriously.
What are the most important aspects of training and education for forensic image and video analysts?
Any manager of a forensic department must look at what the job involves. What are the tasks that need to be performed, and to what standard? This is the starting point. It enables managers to identify what skills their staff need and to recognize where the training should be delivered to fill any skill gaps.
Training should be based on business needs and on competency requirements. It is critical that the outcome of training is that staff become competent at what they do and this needs to be demonstrated by having training that is independently accredited.
We cannot produce a set of validation tests and documents that cover every single possible scenario of how a piece of media might be handled in a forensic workflow.
But we can give the practitioners the tools and knowledge they need to make informed and professional decisions. That way we can trust they will make good judgment decisions based on sound rationale and be confident they have approached each task correctly.
How do you think the world of image and video forensics will change over the next few years?
There will be more video of higher quality which will need to be managed, stored, processed, and presented without adding delays to the workflows. Processing and analyzing big data will continue to be a growing challenge. CCTV image quality will improve bringing great opportunities to make the best use of forensic tools.
I can see facial recognition being a big new area for us. I am optimistic that audio-visual forensics will be taken more seriously as forensic science and will become mainstream.
Recently there have been a few changes in the UK landscape. Digital Evidence Management Systems (DEMS) are being implemented in several forces and the new version of Digital Imaging and Multimedia Procedure (v.3.0) has been published. What challenges does this pose to the use of video evidence, especially in relation to CCTV? How does this affect our daily work with video evidence?
It has always been the case that video/CCTV evidence arrives in many formats. Once under the control of law enforcement, the chain of evidence must begin, and the integrity of the evidence must be maintained. This is true for all evidence, be that a knife, fingerprint lift, or CCTV file.
National guidance for video evidence states an evidential Master must be created using W.O.R.M. (Write Once, Read Many) media such as CD/DVD data disk or a secure server. If another copy of that video is required, a Working Copy can be produced. It is important that the Working Copy remains forensically the same as the Master, bit for bit.
If I alter the nature or content of that file (such as creating a shortened clip of the video, creating a still image from one of the video frames, enhancing the video, or converting it to a different file format) I now have a product which is different. This is not a Working Copy, it is a new derivative, which must be defined as a new Master. That’s the fundamental basic rule.
Now, with that in mind, let’s look at – a scenario. If I take a proprietary CCTV file (.dat as an example) and upload that to a police Digital Evidence Management System (DEMS), I would expect the DEMS to hold a Master copy of the .dat file (and perhaps a Working Copy of it).
A smart DEMS would also recognise the .dat format and transcode it to a playable .mp4 file which can then be viewed in a common video playback app such as Windows Media Player. In doing so it would have created a new Master (the MP4 file is significantly different from the .dat file, so is a new Master). This is also true of practitioners who transcode the video – they should understand that they have changed the content of the file and have now created a new Master.
The problem is that many DEMS systems and many practitioners will transcode the file to create a playable format but store it as a Working Copy of the original, rather than a new Master.
This happens because there is a widespread fundamental misunderstanding about the complexity of audio-visual forensics work. If we are to avoid legal challenges in court the integrity and authenticity of our evidence must be secure. System manufacturers and senior decision makers in this field need to understand the technical challenges, or at least take advice from the technical experts.
How did you learn about Amped Software?
While contributing to a UK national CCTV steering group we recognized that there was no “standard” for AV training.
With the emergence of ISO 17025, we recognized that we needed to set the framework for training and competency. This led me and others to develop the national CCTV competency framework and, as a result, identifying training providers who could deliver high quality training to UK practitioners.
This in turn led me to Amped FIVE as a tool that could provide what we needed in terms of forensic processing and analysis of video files.
Why did you choose Amped Software products?
We use a wide range of tools, depending on the nature of the task at hand. What we do like about Amped FIVE and Amped DVRConv is the close-knit community that exists around them.
There is a strong professional community that shares ideas, suggestions, methods and generally helps each other out. The company is also keen to evolve and develop and in doing so really does listen to its users and takes on board their requirements and ideas for development.
Do you have any interesting stories or success cases related to Amped Software products?
We often find that we can pose a question about a particular CCTV file format (as we know there are over 3,500 CCTV systems in the UK marketplace and that number increases weekly).
Usually, in a very short timeframe, we receive a response back which is either “we’ve had that before, try this solution…” or “send us some details and a sanitized image and we’ll reverse-engineer”.
When you are not busy looking at digital evidence, what do you like doing in your spare time?
I love walking in the hills around Yorkshire (we have five national parks close by) with friends. I also enjoy taking my wife and children on cycling trips. Other than that, depending on the amount of time available I enjoy reading biographies, cinema, restaurants, international travel, photography, and single malt whisky.
I recently testified in court as a forensic image and video expert and, as is sometimes the case, the use of some filters to enhance images was questioned. As I have written before, there is some processing that should be entirely avoided, since it lacks accuracy and repeatability. For example, we should avoid techniques which add new information relying on data obtained by a training set, or techniques which have a random component.
