Using OpenCV to recognize graphic patterns - page 2
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Thanks for the video, I'm hooked : )
But it's not exactly what we need, but it's okay for general education. We need to recognize (memorize, whatever) 2 graphical patterns and compare them for similarity. This is the first task to start with. I'm not sure if we need to train a neural network for that
Obviously you don't quite understand what OpenCV is. It is a library of fast matrix/vector operations running on multiple computer cores or multiple graphics card cores.
Pattern recognition is a field such as machine learning. Deep, convolutional and other special neural networks can be used for this purpose.
The use of OpenCV neural networks in training gives a significant performance boost. That's all.
So your question = cart before the horse.
First you have to figure out how you will define the "pattern" (bitmap image? numeric vector? or something else?).
Learn, learn and learn.
Good luck
You obviously don't quite understand what OpenCV is. It is a library of fast matrix/vector operations performed on several computer cores or many graphics card cores.
Pattern recognition is a field such as machine learning. Deep, convolutional and other special neural networks can be used for this purpose.
The use of OpenCV neural networks in training gives a significant performance boost. That's all.
So your question = cart before the horse.
First you have to figure out how you will define the "pattern" (bitmap image? numeric vector? or something else?).
Learn, learn and learn.
Good luck
Thank you for the unclear comment :)
I'm interested in the accuracy of the definition at this stage, what this thing is capable of at all, without going into deep detail... Can it recognize and compare patterns more accurately than I would through correlation, for example. I don't really care if it's a bitmap or a vector. I understand it already comes with trained layers, and nothing there is no need to train, it will just give a finished result ... but you can also train it for their own purposes, which is already more complicated
Il, can you advise some other method of comparing two curves, more accurate? So that the neural network would not give me something like "yes, I identified this graph, this is a real graph, I'm good... but I can't guarantee the accuracy".
Or application of this method would be limited by intensive training of neural networks, selection of their configurations, selection of training samples, etc... I don't really want to do that for the next 50 years of my life
Thank you for your indecipherable comment :)
I'm interested in the accuracy at this stage, what this thing can do at all, without going into deep details... Can it recognize and compare patterns more accurately than I would through correlation, for example. I don't really care if it's a bitmap or a vector. I understand it already comes with trained layers, and nothing there is no need to train, it will just give a finished result ... but you can also train it for their own purposes, which is already more complicated
Il, can you advise some other method of comparing two curves, more accurate? So that the neural network would not give me something like "yes, I identified this graph, it's a real graph, I'm good... but I can't guarantee the accuracy".
Or the use of this method would be limited by intensive training of neural networks, selection of their configurations, selection of training samples and so on... which I do not really want to do for the remaining 50 years of my life
Here is an example of license plate recognition on Matlab
http://matlab.exponenta.ru/imageprocess/book2/61.php
And other articles on the subject
http://matlab.exponenta.ru/imageprocess/book2/
http://matlab.exponenta.ru/imageprocess/book2/58.php
I wanted to teach the system to see Elliott waves.
Google taught smartphones to recognize speech, I think we can teach them to see waves, too.
I created a similar topic, too.
I wanted to teach the system to see Elliott waves.
Google taught smartphones to recognize speech, it seems to me that it is possible to teach them to see waves.
It is possible without libraries and without NS. The indicator instantly recognizes and numbers up to 9999 patterns. You can do more, but there is no need in such quantity.
It is possible without libraries and without NS. The indicator instantly recognizes and numbers up to 9999 patterns. More is possible, but there is no need in so many.
9999 is nothing compared to different variants of formations, which tend to infinity. It is necessary not only to recognize a pattern, but to recognize any user-defined pattern, any piece of the chart at all, with high precision.
I don't think the price repeats its model with such precision, so I don't bother with such high precision. Who likes it, of course, and who understands it.