Recognition view

Features

You can select (limit) detection features that you would to run for your working document.

Images and keywords detection

This option will perform:

  • searching for exact copies of your project or images similar to yours over the web.
  • analyze the content of your project and suggest the keywords list that might recorded with your image and used later in search by stored metadata.

Text Detection

This will perform a simple text detection on your image. That option is designed for detecting individual words that may be hand written and rotated. May not be accurate for large block of texts or tiny text but on other hand is cheaper and less computing expensive.

Document / dense text OCR

This will perform a dense text detection on your image. That option is more accurate for large block of texts and tiny text but is more expensive in terms of computing and pricing. This option assumes that it will detect machine written text that is inline. Thus any skewed or slightly rotated text will be restored as straightened

You can have selected Text Detection or Document text detection / dense text OCR but not both at same time.  If you are not sure which one is good enough for your case, try to start from Text Detection. If results are not accurate, switch to advanced and more expensive option - Document text detection / dense text OCR.

Example image and results

This exemplary image is used to present the results of the Krasbit Recognition script running at Adobe Photoshop application.

Click Start button to initialize detection process. You will need an internet connection available to process. It may take some seconds to export temporary image, sent it to cloud for analysis and parse response.

Depends of your initial choice of features  - you will get all or only a subset of accordion sections representing analysis results: Keywords, Web images detection, Text detection.

Keywords

Recognition keywords

This section will show you a suggested keywords / descriptions for your image as interactive items. You can click on desired items to be transfer them into document’s keywords. This may help you later to search or filter images by any software capable to work with keywords like Bridge or Lightroom.

Document keywords

This section will show you list of keywords that are already assigned and stored inside your document. You can review them and click on desired items in order to remove them.

Web images detection

This panel shows thumbnails of images found over whole web for which machine learning analysis service classified similarity as one of:

  • Full - represents images that are identified as full copy of your opened document
  • Partial - represents images that are identified as a partial copy of your opened document. It means versions that are cropped, rotated or altered geometrically.
  • Similar - represents images that are found to be similar to your document
Similarity filters

You can use filters and limit displayed results only to desired strength of similarity. Use Full, Partial, Similar checkboxes to control visibility of related thumbnails.

Thumb size

You can control displayed thumbnails size using slider

Opening detected images in hosting application

You may open images in application that hosts Krasbit Recognition extension (Photoshop or Illustrator) in three different ways:

  • Mouse over thumbnail and click Open
  • Double click on thumbnail
  • Drag and drop thumbnail from extension to your workspace

Notice: in case of image url that contains no filetype extension - image will be opened in browser instead because host app could not determine type of image upfront and open it properly.

Opening detected images in web browser

You may open images in system’s default web browser at origin website.

Mouse over thumbnail and click Website

Text Detection

Detected individual words

This list shows each detected word individually. You can click desired on desired words in order to place it into your document at given location and rotation angle.  Your application foreground fill color will be used as color for imported text.

Full detected text

This text area will show you all words detected in your image as single block of selectable text. You can copy/paste desired content and use on your own.

Usage counter

This panel lets you track usage of certain features. Every time you click Start button and receive response from service - your currently selected Recognition features will update(increment) theirs local counters. This may help you determine usage of features per computer in case you are sharing your API access code on multiple desktops.

Reset

Use this button to reset counters to 0 for each Recognition feature.