Apple's transition to its own custom-designed processors, based on ARM architecture and branded as "Apple Silicon," has been a major shift in the company's hardware strategy. The first Apple Silicon chip, the M1, was introduced in November 2020 and has been widely praised for its performance and efficiency.

One of the most notable improvements with Apple Silicon is the significant boost in performance compared to previous Intel-based Macs. The M1 chip, for example, offers significant improvements in CPU and GPU performance, as well as improved power efficiency. This allows for longer battery life and cooler operation of the Macs.

In terms of raw performance, the M1 chip has been shown to outpace Intel-based Macs in most benchmarks. For example, the M1 chip's CPU performance is up to 3.5x faster than the previous generation of Macs, and its GPU performance is up to 6x faster. This makes the M1-based Macs ideal for tasks such as video editing, 3D rendering, and other demanding workloads.

Less measured area of performance is Apple Silicons dedicated neural engines, that should allow for much faster and more efficient machine learning tasks. 

Apple's latest Silicon processors like the M1X and M2 also continues to improve on the M1's neural engine, with new cores and improved performance, making it even more powerful and efficient for machine learning workloads.

These improvements make the M2 Max incredibly performant at text recognition. We took the brand new M2 Max laptop for a test run against the i7 MBP (2018). 


Benchmark 1: Text recognition on 198 images


First test was text recognition on 198 images. They were all loaded on OwlOCR 5.2.2 and OCR all was started. Using revision 1 algorithm, 

the Intel Mac completed the task in 1 minute and 34 seconds

. M2 Max MBP blew it out of the water by taking 


Benchmark 2: Text recognition on a single PDF (525 pages)

Second test was text recognition on a single PDF of 525 pages (the Swift programming language). Here the Intel Mac took 

, good time to get started writing up the results. M2 Max based MBP flew by, taking 

Overall, the performance of the M2 Max is is in a class of its own. In text recognition tests it makes the fairly recent Intel MBP look antique. Another significant benefit here is that during the test the Intel MBP was hotter and noisier while the Apple Silicon machine cruised silently through, making for much improved work environment. It was exciting to see how big of a difference the new technology with its ML cores would make and were not disappointed.

M2 Max proves stunningly fast in text recognition benchmark test

OwlOCR in its previous iterations has supported multiple different ways of interfacing with it: a windowed application, a menubar app and Finder extensions.

In OwlOCR 5, a brand new command line interface (CLI) is provided. CLIs are very powerful for integration as you can call them from practically anywhere. 

Start the app through the terminal by calling the executable inside the app bundle from the terminal:

/Applications/OwlOCR.app/Contents/MacOS/OwlOCR --cli

OwlOCR CLI can be used for many different workflows where at the core is optical character recognition. The application run consists of the following steps.


).
If screenshot is selected you need the select the region of your screen with your mouse that you wish to use. Please note the required screen recording permissions for the app that opens OwlOCR (e.g. terminal, iTerm, Hazel).


Define the output.
By default the plain text is output to stdout. To silence stdout, use 

Define output export.
If you wish to output a searchable PDF file, you can define 

 : shows the OwlOCR app icon in dock for the duration of the run
 

 : silence OwlOCR notifications for the duration of the run
 

 : create the output file even if one exists already, overwriting the old file


Example 1, capture area of screen and get text to stdout:

/Applications/OwlOCR.app/Contents/MacOS/OwlOCR --cli --screenshot

Example 2, convert image to searchable PDF:

/Applications/OwlOCR.app/Contents/MacOS/OwlOCR --cli --input myinputimage.jpg --output mysearhablepdffile.pdf --silent



OwlOCR 5 is now released and can be downloaded from the Mac App Store.

OwlOCR 5 command line interface (CLI) 

Creating a searchable PDF from a scanned paper, photo of a paper, screenshot or any other image content, typically consists of a few steps.

1. Find out what is written in the image and where. This done by a technology called 

2. Create a PDF file where you include the image and an invisible layer of text precisely where the text is in the image.

OwlOCR has provided the tools to do this before, but with version 4.5 we are for the first time including Finder Extensions, a way to do the steps above quickly and easily, right from the Finder.

MacOS requires the user to select and enable the Extensions they wish to see when they are working in their Finder. Fortunately, this is easy to do and you only have to do it once.

System Preferences -> Extensions -> Finder

.
2. Enable the OwlOCR extensions by checking the box for Create Searchable PDF, Extract Text and Create Text File. Or only the ones you wish to use.

Now that the relevant extensions have been enabled you should see them as Quick Actions options in Finder when selecting image and/or PDF files. Please note that for other file types or directories the options are not displayed. 

1. Open a Finder window and navigate to a directory where you have one or more image and/or PDF files.
2. Select the files, right-click and choose 

.
3. The action extension starts processing the files and displays a simple UI in the window to let you know what's happening. The action creates searchable PDFs for each of the files selected.


To see a demo of this in practice check out this video: 



https://www.youtube.com/watch?v=GeT4WqO_htg

How to create searchable PDFs from photos, images and PDF files in MacOS Finder

A whole new way to use the app - Finder Extensions

For the first time OwlOCR actions can be used directly from the Finder. Create searchable PDFs, extract text from files to clipboard or plain text files - with only a couple clicks needed in Finder.

Right-click and select the action from Quick Actions.

The output files will be placed in the same directory and the original files are always preserved.

Please note that the extensions must be enabled in System Preferences > Extensions before they are shown in the Finder. 

OwlOCR now offers the option of including Smart Quotes post-processing. This means replacing quotes with the quote style specified at the system level.

In addition to the previous option of exporting to searchable PDF files, OwlOCR now also supports saving the results to a text file. This can be triggered from the new Finder extensions for image and PDF files or right from the app for any valid inputs.

It's unbelievable that it's been a year already of improving this app. 

The app has grown leaps and bounds, perfecting the features and adding a lot of new things along the way. 

Thank you for all of your support whether that's been a kind email, reporting a bug, sharing the word, providing media publicity or purchasing the Pro license. 

I couldn't do it without you and I wouldn't do it if it wasn't for you. 

OwlOCR v4.5, birthday edition 🎂🎉, released!

The line spacing setting in OwlOCR is a setting that allows you to manage how the OCR output is divided into paragraphs.

In this example I have captured a paragraph of text from Microsoft Word. The text is formatted to have double line spacing (line spacing 2.0) in Word. OwlOCR on the other hand has the default line spacing setting of 1.0. Due to the mismatch in settings, OwlOCR divides almost each line in the output to be a new paragraph resulting in a less than desirable output.

By changing the line spacing setting in OwlOCR to 2.0 and then running the OCR processing again on the same input, the paragraph breaks are corrected to match the input.

If you are getting too few or too many paragraph breaks in your output, it can be a good idea to try to change the linespacing setting to something slightly smaller or larger, even if does not exactly match what you have set in your input document.


Changing line spacing to correctly run OCR text recognition on double spaced documents on the Mac

While in general the OCR engines do a pretty good job these days, none of them are unfortunately perfect. There's always a case where some word or character gets continuously incorrectly detected and one has to go back to fix it.

OwlOCR introduced post-processing rules to help with such scenarios. In the current version, using the post-processing rules, you can define a simple find/replace relation. Simply put, enter an input criteria and what it should be replaced with in the results.

We try to keep things simple, but there are some occasions that can lead to confusion with this one. It helps to understand when the rules get applied.

OwlOCR flow for the text recognition is

User chooses an input file or grabs a screenshot.

User clicks the OCR page or OCR all pages buttons to start OCR.

OwlOCR runs text recognition on the images, noting the locations of objects that look like text and what most probably is written there. These are called observations.

OwlOCR applies the post process rules to each observation separately

. Typically the observations are rows of text as they appear in the input image.

OwlOCR corrects the output by adding line breaks or spaces between the findings as determined by the user's settings.

 the post process rules are applied to each observation separately and before the lines are combined. This means that if you have a post process rule to look for "steaming hot sauna" and in your input image that phrase is split into two lines, it won't be replaced by the post processing rule. This is because when the rule is applied the lines have not yet been combined and as such the full criteria can't be matched. Future versions of OwlOCR may support changing the sequence in which the results are transformed.

Using post processing to improve OCR text recognition results

Images, video and some other content, like slides in Zoom conferences, don't allow selecting and copying text. This is where OwlOCR can help.

Results appear in OwlOCR and are also copied to the clipboard. You may then simply paste your text where you need it (depending on app, typically with 

 is used to extract text from the screen area. The processing is done right on your Mac for near instantaneous results!

If you run into any trouble, be sure to check 

How can I use OCR to capture text from the Mac screen?

Run post-processing steps automatically after each text recognition. In this version we are introducing this feature with simple find and replace steps, but expect to add more advanced use cases in the future. Please send your feedback to guide the direction of this feature if you find it useful.

Some industries have domain-specific lingo that can be tricky for the language correction. Use Custom Dictionary to improve language correction results by adding your words as a comma separated list.

You can now select the folder where instant PDF exports and automatic screenshots are saved.

These new options can be found behind the gear icon in the toolbar or Menubar > Settings. If you need any help getting started, please don't hesitate to get in touch! 

OwlOCR v4.4 released! 

Enable Wrap Naturally to combine rows for naturally wrapping text, disable to get wrapping as in input image. Easy right?

Line spacing can now be selected between 1.0 and 4.0 to better support double spaced documents.

You can now create language sets to support more than one language in a single OCR run.

OwlOCR v4.3 released! 

 the jist by now. 2020. Cringe, shrug and shudder. 

2020 - the year of fabulous missed opportunities. 

Death. Disease. Unemployment. Missed games and events. No Disneyland. Weddings canceled. All bad? 

Chase your dreams because nobody else will 

When California, and other parts of the world, started shutting things down earlier in the year, and today again, everybody had to make a choice. Here is your extra time, what are you going to do with it? 

You can't do what you did before, and time is constant so you're going to spend it in some way anyway. By conscious choice or not.

The popular choice was to use it to buy toilet paper. I tried that. 

Having done that and watched a lot of stuff on Netflix, I chose to put some of that extra free time towards things I’ve always wanted to do. One of those is entrepreneurship or at least building some fun side projects.

Why you can't lose when you're building side projects

I love working on different side projects. Let me break down what's great about them:

1) If you're super successful, you're set for life and you've won. Congrats!

2) If not, you've anyway built something that's worth something and hey, that's a lot better than nothing!  

3) If not, you've built something worthless, but you have probably learned something very useful in the process.

4) If not, let's go have a beer together and thus you've earned a friend!

Since this was talk about side projects, you and your dependants will be fine on your primary income regardless of outcome, so any success will be a bonus, and any loss won't mean you lose your home.

How I found what to build and why you shouldn't do that 

Shame on me, but my side project's inspiration was the anti-thesis of what all the cool kids say about how to get started. 

I tend to watch some of the sessions every year from Apple's WWDC event. In 2019 their Vision framework caught my eye. It was a highly optimised way to run text recognition right on the users' device and seemed to have pretty good results. The idea lingered and over the course of the next half year I increasingly started noticing when and where it could be applied to the Mac (or PC) environment.

The common denominator for the problem was this: there’s 

text on the screen you want to copy, but it cannot be copied.

This happens more often than you realise. 

Maybe the text is on slides in a video or Zoom conference, part of an image, a scanned PDF or a song name in an app like Spotify and the app developer didn’t think that field would get copied.

Solution is easy, right? Just read it and type it like a 15th century scribe.

…but that takes time, I’m a millennial, and the butterfly keyboard on my MacBook Pro is flimsy. Besides, I had learned about that nice first-party API so there was not turning back.

The other technology inspiration was SwiftUI, another cool new framework Apple released the year before that I was dying to try out.

I guess you could say that my idea was born from a personal need and reading up on tech, but if you have a choice, a personal or even better a broader need is the more important thing to consider. 

That is to say, the priority should be 1) what 

Hatching the idea into a shiny beautiful MVP

I had everything I needed. Idea, tech, cool stuff to learn and some time to build it.

I started working on a text recognition app for the Mac with the idea that this is something I'll personally use. I'll be sure to learn a bunch of cool and useful things along the way and that will be a success in itself if nobody else cares.

It only took a couple months to put together the first version of the app. 

The core functionality was that you could simply select an area on your screen and any text in that area is machine read and copied to your clipboard. From there on, you could paste to your notes or emails or whatever is appropriate in your workflow. All of this happens pretty much instantaneously, on-device, so it’s fast enough even for short text snippets and completely private for any business purpose! 

I was actually really happy how well it worked and tried to put extra emphasis to ensure any of my first users would also get a similarly great experience. The way to ensure that is through two key focus points:

. UI should be aesthetically pleasing, but more importantly UX has to be functional and intuitive. I focused on building something that is easy to pick up for anyone, trying to follow good design principles and limiting features in the first version. 

. It's better to have less features that work really well than a lot with bugs. This enables a positive feedback loop where users email you feature requests rather than angry bug reports. That keeps you motivated, your users engaged to your service and gives you healthy backlog where to add stuff.

How it's gone so far and has it been worth it 

Of all the metrics related to apps I value customer satisfaction most. 

The average rating for OwlOCR on the Mac App Store in it's first year so far stands proudly at 

 - roughly doubling in November which was super exciting!

I take monthly active users as some measure of stickiness of the app. I mean, it could be downloaded a gazillion times, but if people don't use it, what's the point? 

But there's a more important thing here beyond the numbers. 

Most rewarding thing overall have been the users that have reached out to express their appreciation of the work. 

In these pandemic times with on and off remote apathy, that has been a great source of inspiration and for those taps on the back I’m truly grateful to my fellow internet strangers.

Has it been worth it? Yes! When there's imposing commands and limitations on your freedom, a fun side project can give you some sense of control. 

You can still work in pursuit of happiness,

 despite the limitations. Side projects can connect you with interesting people, pave way for entrepreneurship or side income and help with isolation depression.

For me and OwlOCR, this year had exciting starting numbers, awesome users and a healthy growth rate, so I'm eagerly looking forward to the next lockdown and 2021 to see if this story can grow bigger!

(And of course hoping for an end to COVID-19. Thank you frontline health workers, you are the real heroes.)

How I turned lockdown into a side project and why you should too

Running text recognition on an image on the screen

This is a very common use case that one can run into daily. There's an image already displayed on your screen and you need to run text recognition on it. 

Running text recognition on an image on the clipboard

If you need to capture all text from an image and it's already on your clipboard or it's most convenient for you to copy it to the clipboard, you can easily input it to OwlOCR.

To use an image from the clipboard in OwlOCR, select 

 from the menubar. After you've input your image, start the text recognition from the toolbar for the currently selected page or for all pages. 

Running OCR text recognition on an image file on the Mac 

There are a couple easy ways to select your image files to OwlOCR.

Drag one or more image file(s) to the OwlOCR window.

 from the toolbar or menubar to browse and select the files.

The images should appear in your OwlOCR. Then simply start the text recognition from the toolbar for the currently selected page or for all pages.

How to run OCR text recognition on an image on the Mac?