Cruise Ship Surveillance Reviewed

Would one option for the man overboard technology be to install cameras to monitor the sides of a ship and to detect events involving a large object, the size of a human being, falling across that area? How many false alarms could that generate?

Obviously, if cameras are targetting the gangways and movement close to the railings, the number of false alarms will be tremendous.

John Honovich

IPVM

Yes, that's the general approach. A few issues we have heard from people familiar with using analytics in this fashion include:

People falling move fast so there is little time to catch the event and a need to make the call quickly.
The position of the cameras can move creating issue with calibration
Wake turbulence can cause false alarms

Alain Bolduc

In reply to John Honovich

Shouldn't targeting a narrower vertical area of the hull prevent the alarms generated by wake turbulence?

How about a faster frame rate to pick up human sized falling objets more accurately? Or even a slower shutter speed - there's been a lot of discussion about that lately? Couldn't a human-sized ghosted streak across the field be used to more easily identify positive events in this particular case?

As far as calibration issues are concerned, there might be some potential if using some kind of specialized enclosure where the cameras would be bolted down to alleviate unwanted movement.

John Honovich

IPVM

I don't have experience to speculate on such detailed cruise specific issues. However, slowing the shutter and making people a blur would likely be bad because lots of others might cause blurs as well. Then how could you tell the difference?

The bigger thing, and here it's just an educated guess, I presume they are going to use thermal cameras because that would help a lot in eliminating the same type of lighting / non-lighting issues that are a pain in general analytics.

Alain Bolduc

In reply to John Honovich

Thermal should definitely be a plus in locking down a solution.

I'm not sure what you had in mind, but I can't think of too many objects that would cause a streak the size of a human being coming down the side of a ship's hull in a vertical drop.

John Honovich

IPVM

A bird. A bath towel dropped off a deck. A flashlight beam.

My general point is that if you let things blur, more things will look a like because all objects become more nebulous.

Barry Shaw

I think the question of using video motion to detect falls in real time was addressed in the report:

"....is not yet reliable in a maritime environment because of the movement of a vessel, weather and sun glare, and lens encrustation caused by saltwater, among other things....a beach towel goes flying over the rail and the man overboard technology picks it up. Or maybe its a sea gull or whatever, but there are a lot of false positives”

I tend to agree.

As for Alain's suggestion "targeting a narrower vertical area of the hull prevent the alarms" - I appreciate where he is coming from, but having installed surveillance on over 85 cruise ships I have to state the obvious - they are really, really, big! The sheer surface area of the exposed (above water) hull would require hundreds, if not thousands, of cameras narrowly focussed on areas through which a human body falling would be analytically distinct enough to trigger an alarm, with enough certainty to warrant investigation. Add to this the fact that the cameras need to be in stainless steel housings and be mounted where they can be regularly cleaned of salt crust and it becomes a very expensive project.

Flir cameras are used on many ships to watch for warm bodies falling overboard. Last weekend in fact, a woman fell off a Carnival Cruise ship and her fall was recorded from the Flir camera, although this was only discovered after she had been missed by friends and later rescued when the ship circled around looking for her.

Here is a YouTube clip of the incident, along with a (claimed) solution to the Man Overboard detection problem.

John Honovich

IPVM

In reply to Barry Shaw

Barry, thanks for the field feedback!

Btw, that youtube video is an excerpt of the same one we have in the report above, in case any readers are curious to know how they relate.

Carl Lindgren

I concur. Although I've never installed on a cruise ship, I've been on four cruises (no, I didn't stay at a Holiday Inn) and checked out their cameras, as I'm wont to do.

I can barely imagine the technical and logistical problems installing any system on a cruise ship, let alone a "man overboard" system sensitive enough to cover everywhere someone is likely to fall off without generating a gargantuan number of false positives. Cruise ships are huge, with multiple outside-accessible locations, including at least 3-4 sets of rails on multiple decks plus all of the balconey staterooms and suites.

They also have "features" that jut out from the sides and stern of the ship and, as you see in the video, lifeboats on both sides. Camera placement would be challenging, as would cabling - I believe cruise ships must have all cabling enclosed in conduit.

This is the ship I'll be sailing on this spring:

How would you cover that? The ship is 965 feet long and 106 feet wide. There are at least 10 decks where someone could jump or fall off. And she is far from the largest cruise ship. RCCL's Oasis class ships are 1,187 feet long by 198 feet wide, with more than a dozen exposed decks:

Carlton Purvis

In reply to Carl Lindgren

You should ask them if they have ever tried out one of those systems during the safety briefing.

Vasiles Kiosses

How about installing a net around the ship to catch falling people. Put a sensor on it based on weight displacment, and then monitor the net with cameras and correlate the events. Probably more reliable no? and they can deploy the nets as they make sail.

Carl Lindgren

Vasiles,

Where would you put them? They can't be placed where they would interfere with the launch of lifeboats. Any higher and the nets would either allow falls from a couple of decks or block balconey staterooms' view.

Vasiles Kiosses

I guess thats the engineers job. I was just thinking out loud. Even in evactuation scenario, how is someone at the lowest deck or the top deck gong to get to the lifeboats in time anyway given the population on the ship. I think it safer just not to go on a cruise.

Barry Shaw

Personally I think it's a little unfair to place the onus on the cruise ship companies to make their decks completely idiot-proof. Decks and balconies are designed to be as safe as possible so as to avoid genuine accidents, but short of caging all around the ship and ruining the experience of the open ocean for everybody, that should be enough.

I'm not aware of any overboard incidents which were truly accidents - in which normal behavior was involved - and I don't consider someone getting rat-faced in the bar and leaning too far over the rails to throw up or simply pass-out, as 'normal behavior' or even an accident in the true sense. Many overboard situations are the result of suicide or homicide - there is dark humor in the industry about a cruise being the preferred way to get rid of your spouse! In those situations the best use of surveillance is on the decks themselves, so that there is evidence of the circumstances in which a person goes over (jump, fall or pushed) and those are the areas which are reasonably well covered by CCTV.

Alain Bolduc

In reply to Barry Shaw

That's all and well if someone goes missing and/or you're trying to reconstruct events after the fact but it won't help the poor sod that ends up in the drink, however that may have happened, which is the whole point of this regulation.

As far as the case at hand is concerned (rat-faced passenger trips over the rail and falls off the boat) I'm wondering what grounds she might have for suing the cruise line? Sounds to me like an attempt at a money grab.

The cruise line should counter-sue for the trouble and expense of having to turn back to fish her out. :D

Barry Shaw

•Jan 23, 2014