I've looked everywhere and yes, including on Tesseract official documentation, but I just couldn't find out how does the page segmentation work? Nor was I able to find the source code, only the declaration of classes & enums.
Specifically, I am interesed in finding out how (the exact steps/ "science") the enum psm=PSM.SINGLE_COLUMN work. What it takes into considerations, how it's adjusted etc.
The reason I need this information is because I chose to implement an OCR application for my final underdrag thesis and my teacher won't accept a "it's an enum and it does things behind I don't know/understand, but it works" answer:)
Thanksss,
Related
I’m new to deep learning and I was reading some state of art papers and I found that mask r-cnn is utterly used in segmentation and classification of images. I would like to apply it to my MSc project but I got some questions that you may be able to answer. I apologize if this isn’t the right place to do it.
First, I would like to know what are the best strategy to get the annotations. It seems kind of labor intensive and I’m not understanding if there is any easy way. Following that, I want to know if you know any annotation tool for mask r-cnn that generates the binary masks that are manually done by the user.
I hope this can turn into a productive and informative thread so any suggestion, experience would be highly appreciated.
Regards
You can use MASK-RCNN, I recommend it, is a two-stage framework, first you can scan the image and generate areas likely contain an object. And the second stage classifies the proposal drawing bounding boxes.
But the two-big question
how to train a model from scratch? And What happens when we want to
train our own dataset?
You can use annotations downloaded from the internet, or you can start creating your own annotations, this takes a lot of time!
You have tools like:
VIA GGC image annotator
http://www.robots.ox.ac.uk/~vgg/software/via/via_demo.html
it's online and you don't have to download any program. It is the one that I recommend you, save the images in a .json file, and so you can use the class of ballons that comes by default in SAMPLES in the framework MASK R-CNN, you would only have to put your json file and your images and to train your dataset.
But there are always more options, you have labellimg which is also used for annotation and is very well known but save the files in xml, you will have to make a few changes to your Class in python. You also have labelme, labelbox, etc.
It seems that nltk.Text.generate() is not available in NLTK 3.0 (see this answer). How should we be generating sentences instead? Thanks.
Unfortunately the generate() function relied on a buggy implementation of ngram models. It has been removed from NLTK 3.0 until someone can get around to fixing it, as you can see here (search for the words "removed ngram model package"). No replacement for this functionality has been provided.
The package nltk.model is still present in the NLTK 3.0 source tree, but it is not part of the distribution. So in principle you could download the source and get it to work, but given the bugs that led to its removal, it's probably a better idea to do without it, or to roll your own. Random text generation is not very interesting unless you control the generation algorithm, anyway.
How to convert from chinese characters to hanyu pinyin?
E.g.
你 --> Nǐ
马 --> Mǎ
More Info:
Either accents or numerical forms of hanyu pinyin are acceptable, the numerical form being my preference.
A Java library is preferred, however, a library in another language that can be put in a wrapper is also OK.
I would like anyone who has personally used such a library before to recommend or comment on it, in terms of its quality/ reliabilitty.
The problem of converting hanzi to pinyin is a fairly difficult one. There are many hanzi characters which have multiple pinyin representations, depending on context. Compare 长大 (pinyin: zhang da) to 长城 (pinyin: chang cheng). For this reason, single-character conversion is often actually useless, unless you have a system that outputs multiple possibilities. There is also the issue of word segmentation, which can affect the pinyin representation as well. Though perhaps you already knew this, I thought it was important to say this.
That said, the Adso Package contains both a segmenter and a probabilistic pinyin annotator, based on the excellent Adso library. It takes a while to get used to though, and may be much larger than you are looking for (I have found in the past that it was a bit too bulky for my needs). Additionally, there doesn't appear to be a public API anywhere, and its C++ ...
For a recent project, because I was working with place names, I simply used the Google Translate API (specifically, the unofficial java port, which, for common nouns at least, usually does a good job of translating to pinyin. The problem is commonly-used alternative transliteration systems, such as "HongKong" for what should be "XiangGang". Given all of this, Google Translate is pretty limited, but it offers a start. I hadn't heard of pinyin4j before, but after playing with it just now, I have found that it is less than optimal--while it outputs a list of potential candidate pinyin romanizations it makes no attempt to statistically determine their likelihood. There is a method to return a single representation, but it will soon be phased out, as it currently only returns the first romanization, not the most likely. Where the program seems to do well is with conversion between romanizations and general configurability.
In short then, the answer may be either any one of these, depending on what you need. Idiosyncratic proper nouns? Google Translate. In need of statistics? Adso. Willing to accept candidate lists without context information? Pinyin4j.
In Python try
from cjklib.characterlookup import CharacterLookup
cjk = CharacterLookup('C')
cjk.getReadingForCharacter(u'北', 'Pinyin')
You would get
['běi', 'bèi']
Disclaimer: I'm the author of that library.
For Java, I'd try the pinyin4j library
As mentioned in other answers the conversion is fuzzy and even google translate apparently gets a certain percentage of character combinations wrong.
A reasonable result which will not be 100% accurate can be achieved with open-source libraries available for some programming languages.
The simplest code to do the conversion with python with the pypinyin library (to install it use pip3 install pypinyin):
from pypinyin import pinyin
def to_pinyin(chin):
return ' '.join([seg[0] for seg in pinyin(chin)])
print(to_pinyin('好久不见'))
# OUTPUT: hǎo jiǔ bú jiàn
NOTE: The pinyin method from the module returns a list of possible candidate segments, and the to_pinyin method takes the first variant whenever more than one conversion is available. For tricky corner cases this is likely to produce incorrect results, but generally you'll probably get at least a ~90..95% success rate.
There are a few other python libraries for pinyin conversion but in my tests they proved to have a higher error rate than pypinyin. Also, they don't appear to be actively maintained.
If you need better accuracy then you'll need a more complex approach that will rely on bigger datasets and possibly some machine learning.
Is there a good book or other references on Text User Interface Design? I am not interested in graphical user interfaces. I am interested in usability for good command line and scripting interfaces.
Your interface should follow the Rule of Least Surprise as described by ESR in The Art of Unix Programming. If your programm supports command line options, make sure they have the traditional meaning. Be sure to read the chapter about Tradeoffs between CLI and Visual Interfaces.
IBM developed a standard called Common User Access. The Common User Access Basic Interface Design Guide has been published in the BookManager format and in HTML here.
The guide was written as a standard for developing 3270 applications. In my opinion the most important parts are the function keys standard and a color standard.
I'd use a favorite program as a reference for something like this. What command line utility do you think has a good, efficient interface that you could model your program on? Use it.
Update: So I think I need to revise this a little. It was taken way too literally. Google and this site proved that the internet is very democratic. What is popular is replicated, linked to or reproduced in someway.
Given this, plus one's personal experiences with computers, I think it is feasible to derive a pretty good solution based on personal experience and consideration for the solution to be provided.
For example, vim is a great program. A lot of people use it and love it. But that type of interface is probably not going to work (at least well) for a version control system. But both interfaces are very elegant for the purpose they suite. On the other hand, the vim type interface might work for a section of the version control system -- the commit dialog for example.
Now, I know that vim is normally used for the "commit dialog" (by default) for svn (on unix based OSes). This is just an example of mixing two styles of interfaces to come up with a cohesive solution.
You should have a look at some of the ideas behind Ubiquity as well as some of the ideas Aza Raskin talks about, seems like the same kind of thing.
How does one study open-source libraries code, particularly standard libraries?
The code base is often vast and hard to navigate. How to find some function or class definition?
Do I search through downloaded source files?
Do I need cvs/svn for that?
Maybe web-search?
Should I just know the structure of the standard library?
Is there any reference on it?
Or do some IDEs have such features? Or some other tools?
How to do it effectively without one?
What are the best practices of doing this in any open-source libraries?
Is there any convention of how are sources manipulated on Linux/Unix systems?
What are the differences for specific programming languages?
Broad presentation of the subject is highly encouraged.
I mark this 'community wiki' so everyone can rephrase and expand my awkward formulations!
Update: Probably didn't express the problem clear enough. What I want to, is to view just the source code of some specific library class or function. And the problem is mostly about work organization and usability - how do I navigate in the huge pile of sources to find the thing, maybe there are specific tools or approaches? It feels like there should've long existed some solution(s) for that.
One thing to note is that standard libraries are sometimes (often?) optimized more than is good for most production code.
Because they are widely used, they have to perform well over a wide variety of conditions, and may be full of clever tricks and special logic for corner cases.
Maybe they are not the best thing to study as a beginner.
Just a thought.
Well, I think that it's insane to just site down and read a library's code. My approach is to search whenever I come across the need to implement something by myself and then study the way that it's implemented in those libraries.
And there's also allot of projects/libraries with excellent documentation, which I find more important to read than the code. In Unix based systems you often find valuable information in the man pages.
Wow, that's a big question.
The short answer: it depends.
The long answer:
Some libraries provide documentation while others don't. Standard libraries are usually pretty well documented, whether your chosen implementation of the library includes documentation or not. For instance you may have found an implementation of the c standard library without documentation but the c standard has been around long enough that there are hundreds of good reference books available. Documentation with hyperlinks is a very useful way to learn a new API. In any case the first place I would look is the library's main website
For less well known libraries lacking documentation I find two different approaches very helpful.
First is a doc generator. Nearly every language I know of has one. It basically parses an source tree and creates documentation (usually as html or xml) which can be used to learn a library. Some use specially formatted comments in the code to create more complete documentation. JavaDoc is one good example of this. Doc generators for many other languages borrow from JavaDoc.
Second an IDE with a class browser. These act as a sort of on the fly documentation. Some display just the library's interface. Other's include description comments from the library's source.
Both of these will require access to the libraries source (which will come in handy if you intend actually use a library).
Many of these tools and techniques work equally well for closed/proprietary libraries.
The standard Java libraries' source code is available. For a beginning Java programmer these can be a great read. Especially the Collections framework is a good place to start. Take for instance the implementation of ArrayList and learn how you can implement a resizeable array in Java. Most of the source has even useful comments.
The best parts to read are probably whose purpose you can understand immediately. Start with the easy pieces and try to follow all the steps that are hidden behind that single call you make from your own code.
Something I do from time to time :
apt-get source foo
Then new C++ project (or whatever) in Eclipse and import.
=> Wow ! Browsable ! (use F3)