Any recommended OCR software for Linux?

[rcentros]

Originally Posted by Pajamaman

I agree totally; however, this thread began talking about, and I continued to talk about Linux distros like Ubuntu for example you install on a desktop. Clearly Android and all the other derivutives etc are not the same thing. We are talking about desktop pc systems and pc Linux. You are changing the goal posts.

I'll grant that Android (though built on the Linux kernel) is not a Linux distribution, but that argument doesn't really apply to ChromeOS, a GNU/Linux distribution based on Gentoo Linux. Basically a Chromebook is Linux running the Chrome browser. It can now run Android applications (using a virtual machine) and it also runs Linux applications (in a container). And Chromebooks have been consistently outselling Macs for the last year, or year and a half. They are becoming a significant part of the "desktop" world (desktop referring to desktops and/or laptops).

[Sarmat89]

Originally Posted by pazos

Nowadays that doesn't hold true and Unix-like systems are meant for end users. You can ask millions of Android/iOS/mac/chromebook users.

Linux works for people who are OK with using obsolete technology: supercomputers of '50s, servers of '70s, primitive X-based GUI apps from '80s, or is content with a walled gardern environment which is completely OS-agnostic (like Android and MacOS, as well as various embedded uses).

For any creative work outside of vendor-locked apps you need a real OS, one that doesn't require you to two-pass compile your text documents, and doesn't treat your screen or console as being located on another continent.

I'm sure tesseract would work well if you did your own geometry correction, binarization, despecling, segmentation for it. Unsurprizingly, the primitive state of Linux software makes creating such a tool impossible.

[salamanderjuice]

Originally Posted by Sarmat89

Linux works for people who are OK with using obsolete technology: supercomputers of '50s, servers of '70s, primitive X-based GUI apps from '80s, or is content with a walled gardern environment which is completely OS-agnostic (like Android and MacOS, as well as various embedded uses).

For any creative work outside of vendor-locked apps you need a real OS, one that doesn't require you to two-pass compile your text documents, and doesn't treat your screen or console as being located on another continent.

I'm sure tesseract would work well if you did your own geometry correction, binarization, despecling, segmentation for it. Unsurprizingly, the primitive state of Linux software makes creating such a tool impossible.

This is some trolling.

There's nothing primitive about Linux or FOSS software. You don't need to two pass compile a document in LibreOffice on Linux anymore than you do Word on Windows. People use LaTeX on Windows and Mac too because it's still the gold standard for typesetting math and handling things like references automatically. Doing that in Word is literally PAINFUL. Even though Word is supposed to support LaTeX syntax it barely works and takes five times as long to have dig through the equation editor to find what you need and it STILL can't automatically number equations. Have fun adjusting all your equation numbers (including where they are referenced in the text!) when you need to insert one in the middle of a document. I will take waiting an extra 5 seconds to compile a TeX document twice to get references right vs. 30 minutes to adjust and find every equation number.

There's absolutely no reason something like FineReader couldn't work perfectly fine on Linux if they cared to port it. Zero. It's also possible to run a Linux GUI without using X at all anymore. Wayland has been a thing for several years at this point. You need to update your worldview which seems to be stuck in 1994 or something.

[Sarmat89]

Originally Posted by salamanderjuice

five times as long to have dig through the equation editor to find what you need and it STILL can't automatically number equations.

MS Word for Windows could automatically number everything since version 1 (late '80s). Equation editor is a freebie toy for schoolchildren, there is a full version for professional use.

There's absolutely no reason something like FineReader couldn't work perfectly fine on Linux if they cared to port it.

So where are those applications ported? Why does Apple develop and support three or 4 toolkits to build applications on MacOS instead of using the Linux' godly software?

Why Linux get only rejects like LibreOffice (a rebranding of German abandonware office suite called StarOffice) or Firefox (a rebranding of abandonware Netscape Navigator), or Tesseract (an abandonware Hewlett-Packard product), or Java/Electron 'apps'? Distros are unable to continue to maintain the modern software like Chrome[ium], starting to drop it, why?

Even Cuneiform, a toy-like OCR program bundled with Lexmark hardware, has better UI and functionality than any tesseract frontend... again, why?

[salamanderjuice]

Originally Posted by Sarmat89

MS Word for Windows could automatically number everything since version 1 (late '80s). Equation editor is a freebie toy for schoolchildren, there is a full version for professional use.

Show Steps. Show me how to automatically number equations in Word 1.0 for Windows. And how to cross reference it in the text automatically. No hacky workarounds like stuffing it in an invisible table and referencing that please.

Originally Posted by Sarmat89

So where are those applications ported? Why does Apple develop and support three or 4 toolkits to build applications on MacOS instead of using the Linux' godly software?

Why Linux get only rejects like LibreOffice (a rebranding of German abandonware office suite called StarOffice) or Firefox (a rebranding of abandonware Netscape Navigator), or Tesseract (an abandonware Hewlett-Packard product), or Java/Electron 'apps'? Distros are unable to continue to maintain the modern software like Chrome[ium], starting to drop it, why?

Even Cuneiform, a toy-like OCR program bundled with Lexmark hardware, has better UI and functionality than any tesseract frontend... again, why?

Apple took KHTML the rendering engine designed for KDE's Konqueror web browser and forked it to create Webkit the rendering engine they still use for Safari to this day. Google also forked Webkit to create Blink, the rendering engine used in Google Chrome and Electron apps. All of them owe their start to a piece of software originally designed for Linux.

And if a program is being actively maintained with a new name it's not abandonware, nor are programs abandoned when made open source.

[Sarmat89]

Originally Posted by salamanderjuice

Show Steps. Show me how to automatically number equations in Word 1.0 for Windows.

Create an autotext containing a SEQ field and the equation placeholder, expand it, edit the equation, add a bookmark describing it, refer that bookmark in your text...

[rcentros]

Originally Posted by Sarmat89

Linux works for people who are OK with using obsolete technology: supercomputers of '50s, servers of '70s, primitive X-based GUI apps from '80s, or is content with a walled gardern environment which is completely OS-agnostic (like Android and MacOS, as well as various embedded uses).

I'm trying to understand why you're so bitter about Linux. I don't particularly like Windows or Macs but I'm happy for others who use and like them.

So a modern supercomputer is a mainframe from the '50s? Well I couldn't find specs for a '50s mainframe, but I did find the specs for a '70s one.

Here's an article about a state of the art, 1970s mainframe, an IBM System/370 Model 145...

IBM maintains an awesome archive of its history, collecting various documents and media it's released since being founded in 1911.

We were especially intrigued by this 1970 press release singing the praises of System/370 Model 145, a mainframe computer that was state of the art at the time of its inception. It had 500 KB of RAM, 233 megabytes of hard disk space, and ran at 2.5 MHz. It took up an entire room.

Nowadays, this computer would be able to store a small collection of photos and (slowly) access them. Not much else! You could double its 32,000 characters of control storage to 64,000 by "using a portion of main memory, if needed, to accommodate optionally available functions."

Adjusted for inflation, this computer would cost you between $4.3 million and $10.8 million in today's dollars, depending on the options you selected.

https://www.businessinsider.com/ibm-...s-today-2014-5

So you can see why it's hard to take what you say about Linux seriously. You undercut your own credibility by comparing '50s mainframes with modern supercomputers.

In comparison here are the specs for a modern supercomputer (not the same as mainframe)...

According to a quote with several origins, science advances on the shoulders of giants. In our time, these words have taken on a special meaning thanks to a new class of giants—supercomputers—which nowadays are pushing the boundaries of science to levels that the human intellect would be incapable of reaching on its own.

In a few decades, the strength of these giants has multiplied dramatically: in 1985 the world’s most powerful supercomputer, Cray-2, could process 1.9 billion floating point operations per second (FLOPS), or 1.9 gigaflops, the parameter used to measure the power of these machines. By comparison, a current PlayStation 4 game console reaches 1.84 teraflops, almost a thousand times more. Today, there are at least 500 supercomputers in the world that can exceed a petaflop, or one billion flops, according to the TOP500 list drawn up by experts from the Lawrence Livermore National Laboratory and the universities of Mannheim (Germany) and Tennessee (USA). ...

1. Summit, Oak Ridge National Laboratory (USA)

The world’s most powerful supercomputer today is Summit, built by IBM for the U.S. Department of Energy’s Oak Ridge National Laboratory in Tennessee. It occupies the equivalent of two basketball courts and achieves an impressive 148.6 petaflops thanks to its 2.41 million cores.

https://www.bbvaopenmind.com/en/tech...ntific-giants/

All 500 of the top supercomputers run on Linux. A few years ago there were still a few Windows, Mac and UNIX supercomputers in that mix. There's a reason for that, Linux scales better than the other OS's. That's also why Microsoft (themselves) are moving to Linux servers in their cloud.

The reason there aren't a lot of big, commercial applications for Linux is that the market isn't there. It has nothing to do with the capabilities of the Linux operating system.

I don't care why you seem to hate Linux, but at least try to stay grounded in reality when write about it.

[gmw]

[delete]

[Greg Anos]

Back to the original OP.

If you need windows software, buy a cheap windows box as a dedicated machine (sterile - not connected to anything external) and a KMV switch.

I build a refurb for less than $200 dollars, and used my retail version of Win 7.

A screaming speed demon? No, an i3 Sandy Bridge Intel NUC. Works just fine for the dedicated use I built it for. OCR is not high frame rate shooter games. More than enough power, and had USB3 for interfacing to scanners/printers.

Double toggle from one machine to another (from the keyboard). No sweat (and 100% compatibility.)

[Sarmat89]

Originally Posted by rcentros

All 500 of the top supercomputers run on Linux. A few years ago there were still a few Windows, Mac and UNIX supercomputers in that mix.

All supercomputers are performing very primitive batch jobs: open file, read data, calculate something, open another file, and write data. Multiplication of 1,000,000,000 of 1,000,000x1,000,000 matrices is as trivial as 2 5x5 ones. Since '80s computers become used for more sophisticated tasks, and UNIX/Linux missed the train.

[rcentros]

Originally Posted by Sarmat89

All supercomputers are performing very primitive batch jobs: open file, read data, calculate something, open another file, and write data. Multiplication of 1,000,000,000 of 1,000,000x1,000,000 matrices is as trivial as 2 5x5 ones. Since '80s computers become used for more sophisticated tasks, and UNIX/Linux missed the train.

You do realize that, ultimately, all digital computers work with binary code, right? All digital computers, no matter how primitive or modern and no matter how fast or slow, ultimately work with zeros and ones (like off and on switches). The computer with the most computing power is the one that can do the most work per second. That means supercomputers running on Linux are vastly superior to your PC. There's a reason supercomputers use Linux, it's called scalability. You can't do what Linux supercomputers do with Windows. Just won't happen. And you don't write word processing applications for supercomputers because that's an extreme waste of their power.

But this too ridiculous to go on. I'll try to let you have the last word.

[orebmur]

Is there no moderator to cut down all those off topic discussions!?

[kacir]

Originally Posted by Sarmat89

...and all that Linux demagogy to somehow defend a plain-text only stock-word recognition with a batch-mode-only software from 1985...

Originally Posted by Sarmat89

Even Cuneiform, a toy-like OCR program bundled with Lexmark hardware, has better UI and functionality than any tesseract frontend... again, why?

Because Tesseract was developed by HP and later on by Google to handle recognition in a non-GUI, batch-mode, automated manner. Like reading all the books scanned by Google to enable us to search them, or scanning street images for street names, signs and suchlike.

As I stated in one of my previous posts, Tesseract works *for me* in a satisfactory manner, and I do not have to buy [or pirate] a commercial software with much nicer GUI for an occasional OCR of a book (that I might want to abandon after reading a few pages). At work I use commercial OCR software, but I need to do completely different things there. And that commercial software has earned its price after a few uses.

Disclaimer: you have my special permission to like or dislike any operating system or software you wish ;-)

[kacir]

Originally Posted by Pajamaman

So is Mint the same as Ubuntu, but just better (i.e. equal driver support)?
...
Mint sounds a bit like Ubuntu+. How is it better than Ubuntu?

I personally find it more usable and likeable "out of box". Nice functional window environment that doesn't require [much] tweaking for my use, most of software and repositories I need available by default, it is based on Ubuntu, which is based on Debian, so you have all that software or knowledge-base available if you need anything special.

... Oh ... one more thing ... it doesn't come with Unity environment or other super-radical stuff that you can't easily change to your liking ;-)

[salamanderjuice]

Originally Posted by kacir

I personally find it more usable and likeable "out of box". Nice functional window environment that doesn't require [much] tweaking for my use, most of software and repositories I need available by default, it is based on Ubuntu, which is based on Debian, so you have all that software or knowledge-base available if you need anything special.

... Oh ... one more thing ... it doesn't come with Unity environment or other super-radical stuff that you can't easily change to your liking ;-)

Ubuntu doesn't come with Unity anymore either. They switched to GNOME 3 with a couple preinstalled extensions like the dock with 17.10. If you don't like GNOME 3 there is KDE and XFCE versions readily available. There's even a cinnamon remix if you want that. You can also just change anything you want...

Last time I used Linux Mint they were forcing DDG as the default browser search engine and they were extremely aggressive about that as it doesn't even come with the option to switch to Google pre-included. It's honestly easier to switch from Bing to Google in Microsoft Edge than Firefox on Mint. At least last time I tried. Left a bad taste in my mouth.