Pool swim

I have the opposite problem - it always under counts.  That could be because I have a short pool at home - actually shorted than the minimum you can configure in the watch.  I would say my FR 255 is better than my FR 245 was in this regard - perhaps because it has a compass?

whogrant said:

perhaps because it has a compass?

Unfortunately, you are wrong twice:

245 has a compass , in addition, the compass is not used when determining turns

whogrant said:

Are they using the accelerometer or just a change in stroke interval as you turn and kick off the end of the pool?

As far as I understand, the watch uses an accelerometer to determine strokes and the interval between strokes, and subsequently turns.

Thanks, I found another comment from someone who was having problems with turn detection and wrote "I talked with two Garmin techs. They said that laps are counted by detecting the pushes against the wall, or more precisely the slowdown, the push in the opposite direction and the glide."

See: https://forums.garmin.com/sports-fitness/running-multisport/f/garmin-swim-2/276511/new-watch-latest-software-swimming-is-barely-recognized-and-turns-off-when-i-turn/1327810#1327810

I've turned off auto rest detect since I read that my swims often have multiple intervals even though I never stopped swimming.  When I do I always manually pause and resume the activity - that may still count as an interval but I know there are swims when I never stopped that had several intervals. I'll see if that changes the behavior - although for me I think its good enough considering my short pool length, I just focus on time of swimming and heart rate mainly - most of my exercise is on the treadmill, or running and hiking outdoors.

whogrant said:

or more precisely the slowdown, the push in the opposite direction and the glide

Thanks, this is useful.

Well, I wonder if those guys swim with the watch themself. I always get overcounting with ridicouslous fast lanes if I look at the watch anywhere else than at turns.

I think the algorithm could be easily improved by using time/speed and compass.

3878047 said:

I think the algorithm could be easily improved by using time/speed and compass.

I don't think a compass is useful for determining turns unless you swim with your watch hand outstretched and are interested in counting strokes. The use of time and wrist speed are unclear to me (the watch is located on it, after all). The existing algorithm, in my opinion, is very smart.

I usually get a few overcountings. In most cases the "wrong" turn ist detected less than 20s from/to a "correct" turn (50m lane). In those cases it is clear that one of the detections is wrong and the algorithm could be improved to select only one.

The compass will generate a pattern depending on style (as the gyrometer does) and direction. This can be used to make better decisions as well.

If Garmin lets us correct wrong turns, they could get additional data on their lap counting performance. 

3878047 said:

The compass will generate a pattern depending on style (as the gyrometer does) and direction.

Sorry, I absolutely do not understand the meaning of this sentence (in my opinion it is missing).