Using OpenCV to recognize graphic patterns
It's not enough to recognize them (patterns), even a human can do it.)
You also need to classify them and understand what to do with them
Everyone knows that the correlation method and similar methods do not exactly cope with determining the correspondence of time series, and in some cases are not accurate at all.
Recently, computer vision has become widespread. Basically, it is used to recognize images, such as faces in photos. As far as I know, this method works very accurately. Does anyone have any experience using these libraries for pattern recognition? Well, and using the libraries in mql. I think the theme is very interesting in itself and should be developed. I don't have any experience with it yet, but I'd like to learn.
I guess it could be great for machine learning, pattern searching and other tasks.
The fun - you can build in a bot possibility to determine the user by his face, if there is a camera, and if it is a different person then do not allow to trade:)
Pruf http://opencv.org/
I propose then to go further in thinking, let's imagine that the recognition mechanism already exists and it works in the terminal, through the same OpenCV.
What next?
Waves of Elliott and Wolf? We have detected them with some accuracy and what next? Then we go back to robotics with SL, TP, TS
I propose then to go further in thinking, let's imagine that the recognition mechanism already exists and it works in the terminal, through the same OpenCV.
What next?
Waves of Elliott and Wolf? We have detected them with some accuracy and what next? Then we go back to robotics with SL, TP, TS
At least a clear and stable identification of "horrible" places for strategies. For example, the sooner something whistles "like that" the better, you can stop the counter-trend stratum and avoid losses.
A simple scenario - mark the loss-making zones on a historical chart (the developer knows them "by sight"), run the tutorial, and the same OpenCV is staring into the chart nonstop.
But you can't rewrite it... out-of-the-box pattern recognition is not designed for graphs. There is still a lot of mathematics to be done.
I propose then to go further in thinking, let's imagine that the recognition mechanism already exists and it works in the terminal, through the same OpenCV.
What next?
Elliott and Wolf waves? We have detected them with some accuracy and what next? Then we go back to robotics with SL, TP, TS
I have already written that this topic is narrowly focused, let's please do not litter it with leftovers, because as usual, you can not find anything on the topic because of people like you, who are ahead of the horse. Application options are varied and beyond the scope of the topic.
If you have something specific to this library, go ahead.
I have already written that the topic here is narrowly focused, let's please do not litter it with leftovers, because as usual you can not find anything on the topic because of people like you, who run ahead of the horses. Application options are varied and beyond the scope of the topic.
If you have something specific on this library - go ahead
And you yourself are already working with this package? Downloaded it, unzipped it, it looks like a monster. There are a lot of printed books on the site, one of O'Raily's is over a thousand pages long!
If you are working, with VS and what version? Or with something else?
There is a doc on the site, I will read it slowly.
Are you already working with this package yourself? Downloaded it, unzipped it, it looks like a monster. There are a lot of printed books on the site, one of the O'Raily over a thousand pages!
If you are working, with VS and what version? Or with something else?
There's a doc on the site, I'll be reading slowly.
I am still looking for the right side of this monster :) I am looking for people who have already worked with this
I need to properly formulate the sequence of steps to implement, for example, a comparison of two patterns, and then do something
The greatest advance in this direction is obtained with CNN (Coiled Neural Networks).
- habrahabr.ru
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. That's the first task to start with. I'm not sure if you need to train neural network for that
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Everyone knows that the correlation method and similar methods do not exactly cope with determining the correspondence of time series, and in some cases are not accurate at all.
Recently, computer vision has become widespread. Basically, it is used to recognize images, such as faces in photos. As far as I know, this method works very accurately. Does anyone have any experience using these libraries for pattern recognition? Well, and the use of libraries in mql. I think the theme is very interesting in itself and should be developed. I don't have any experience with it yet, but I'd like to learn.
I guess it could be great for machine learning, pattern searching and other tasks.
The fun - you can build in a bot possibility to detect a user by his face, if he has a camera, and if it is a different person then do not allow to trade:)
Pruf http://opencv.org/