Personally, i wouldn’t use the word bad, although i am sure there is bad…but, we already know what efficient looks like, so i guess that’s good. Everything else is not as efficient as it could be, and the objective should be to make each runner as close to efficient as they can get…however, almost certainly they won’t…
AI can be trained to the outcome, so if an individuals performance improves correlated to the data it’s analysing, it can find the needles in the haystack that are those correlation points and recommend whatever is required to achieve that.
All of which is to say, no you don’t need a model of running to aim everyone at. In fact we’re more likely to see revolutions by not forcing pre-existing notions into the AI models. Of course, that does make it much harder to be productive and acceptable.
1 Like
Can you see that from my posts? I can share more…
I did the two recommended exercises today, the swimmers were harder than I had expected.
Tried to carry the hands higher during my trail run.
But thats how AI models train, its binary classisication, or linear if using regression model, you can have a numeric range of 0 and 1 or 1 to 10 or 1 to 100 of how good the training data “runner” is, it learns from that. Based on whatever inputs from the runner you decide to train it on.
What you question is exactly why current AI models are not likely to work at this level yet
why not?
1 Like
yes please…
What is over striding?
What is a back position?
Do they mean your recovering leg as back knee angle?
What is good about your front knee angle at foot landing?
1 Like
…Are you thinking this is my app?
Here you go:
oh, and it seems to me like they are comparing you to some model or other…
Well they are.
That doesn’t mean AI has to. Different discussions.
What is the target?
Unsupervised learning, or clustering is with unlabelled data. Im not sure that will work with image based inputs. Images are processed with neural networks currently.
Back to point one, what is the target then
There is a massive misunderstanding of how intelligent current models are, yes they are good at what they do, within reason, but they are far from intelligent enough to reason.
1 Like
thanks - i haven;t time to reaserch it right now…just interested in what it is regsitering…and that it has already displayed an error…
for sure it doesn’t have to…but what direction will it take you in if it doesn’t know which direction to go?
Probably better to move that off into the AI thread.
We don’t think Ochy uses AI so we think it’s an algorithm based on the prevailing ideas coaches hold of what is good and bad.
No worries - and I’ll ask my club coach what he does with it too.
The idea is to regularly upload 10s clips so it gets a better idea of how you run.
The output is a series of exercises, so I’ve got the two upthread, the next week it changes to something else.
It also has recommended warmup drills
can you link to the clip, please?
Don’t think I can, might need to upload to YouTube and send you a link
1 Like
coaches?
Yep…
1 Like
Coaches can analyze and add their own recommendations to the athletes