It appears that Seeing Machines (AIM: SEE) is making good progress in bringing its world-leading, eye-tracking technology products to a variety of transport markets.

Re. today’s news that one of the world’s leading contract manufacturers has taken a 12% stake in Seeing Machines, investing A$12.8m (£6.7m) for 129.7m shares at 5.2p, a 20% premium to the recent share price, finnCap analyst Lorne Daniel commented: “In our opinion, VSI, as well as providing as a source of finance, offers a low-cost development and manufacturing partner for the road-going and other devices.”

Following on my previous interview with Ken Kroeger, I also wanted to add some interesting snippets from last Friday’s interview that might be of interest to those investing (or thinking of investing) in the company.

Fleet

Seeing Machines has started designing the next generation fleet product (which it appears will be manufactured by VSI). It will not only be better than previous iteration (with a forward facing camera) but is expected to be about 40-45% cheaper.

In addition, Ken Kroeger revealed: “We are talking to 8 or 9 of the biggest telematics companies in the world now and getting quite a bit of interest from them.”

Asked whether the deal was going to be exclusive or non-exclusive, he replied: “It will be non-exclusive. I think we will have to offer some differentation; maybe it will be region by region. A lot of these companies have 400,000 – 500,000 units under management.”

As to the product Seeing Machines would offer them: “This next generation will remove all the things that the telematics companies have: they all have GPS, telecomms, power. So we are building more of a partner unit that will sit beside the telematics unit and only provide the services that it has to have as opposed to all the services inside. Again offering a lower cost product that will act as a companion to the telematics product.”

In terms of how this business model will operate, he explained: “I think where this is going, we will start looking at more channel type relationships, looking at our own business model almost like software as a service where they get a piece of hardware, pretty much like a mobile phone deal where you pay something for this low cost unit, it is installed and then we are scraping more of a monthly payment – parallel to the telematics model.”

Rail

Not only has a third trial just started on the railway side but Seeing Machines has also submitted a tender to the Transport Authority at a big US city for a safety solution for its commuter trains.

If successful, it will garner a lot of publicity and Ken believes: “It would really launch us into that rail space.”

Fortunately, the improved algorithms resulting from the auto development mean that SEE’s product doesn’t need a lot of re-engineering to be used for rail, thus reducing the cost and time of deploying it. As Kroeger explained: “It re-captures the faces now very quickly. The old mining technology, our previous set of algorithms, took 15-30 seconds to find and lock onto the face, whereas it now takes less than a second. So you can move away and come back without it losing its effectiveness.”

Indeed, its continually improving its algorithms, as Kroeger revealed: “One of the biggest changes inside the business is that there is this new science called Machine Learning. Instead of writing software to do something you write software that can learn as you feed it new information. So we started doing that about a year and a half ago.

“It was as part of a continual push to improve those algorithms, not only for performance but also in the automotive space you have to deliver them on cheaper and cheaper platforms. You have to continually drive your prices down, so in order to do that you go to cheaper and cheaper processing. You have to keep on improving them.”

I had been concerned whether Seeing Machines could maintain its technological lead in this area but it seems that it has the ability to maintain this ‘moat’ around its business.

Again, Kroeger enthused: “What makes us special, why it is so perfect for us is that there is no other company in the world where, literally we walk into the office in the morning and there are thousands of hours of video captured the day before of drivers. We take that information and it goes through a truthing process, where we have people looking at the video very very closely. They identify where people had a fatigue event and they can annotate that video to highlight key parts of the video. They can look at 1 minute before, 10 minutes before, 1 hour before and deep learning starts to look for tell-tale signs that are common across all users to develop a more predictive algorithm.”

The writer holds stock in Seeing Machines