For Windows, unzip the full version to any folder and run ImageTrans.exe. For Mac, drag the app in dmg file to Application.
Cross Platform Version¶
Download ImageTrans’s zip file, unzip it to any folder, double-click ImageTrans.jar or enter the command line
java -jar ImageTrans.jar to run it.
Configuration of OCR and Machine Translation¶
ImageTrans has integrated common online OCR and machine translation APIs. Usually, you need to apply for their API keys to call them. ImageTrans has included the API keys of some services, which can be used directly:
OCR: Baidu, OCRSPACE, Azure
Machine translation: Baidu, Niutrans, Tencent, Google without api key
In addition, some offline OCR and machine translation are also supported.
If you need to use Tesseract for OCR, please install it by yourself （Link）and specify the path of Tesseract in ImageTrans.
Here is a Windows version for download: Baidu Netdisk (code: ktpt)). Put the downloaded tesseract-ocr with ImageTrans. Put additional language data files under
- The built-in OCR in Windows 10
Windows 10 comes with built-in OCR, but you need to install the language environment first. Actually, it has the same engine as Azure and OCRSPACE. This OCR engine in ImageTrans is called WinRT because it is based on the Windows Runtime API.
- PaddleOCR, EasyOCR
You need to install Python and the above software and run them with the server scripts provided, like the server of PaddleOCR.
It supports calling ABBYY Finereader for OCR. You need to specify the path of FineCMD.exe in the Preferences.
Offline machine translation:
- OPUS-CAT. Opus-CAT is the offline machine translation engine by Helsinki Natural Language Processing Group. Download and install it from here and install the Fiskmo machine translation plug-in to use it.
- eztrans xp. This is a Japanese-Korean translation software. You can learn about its usage in this issue.
Each time you run ImageTrans, a validator will show up. You need to verify with your the email and order number. The order number can be found in the email you receieved after your purchase.
Each email can be used for verification on at most three devices. If you want to use it on the fourth device, you have to reset with your email.
Click File->New, select a place and enter the project name to save the project file.
Click File->Import pictures (from a folder) and select in which folder the picture exists to import. It will read all the subfolders and import existing jpg, png files.
You can also use the context menu to paste a image.
It can also import PDF files and extract selectable text.
You can mark text areas in boxes and get their text. The tool supports manual selection and four automatic selection. Further adjustment is also supported.
Create Text Area Boxes Manually¶
Double-click on the picture to create a selection box. Press on the middle area to move the box and press on the lower right corner to resize.
There is also a quick creation mode. Click the button on the left and you can press and drag your mouse cursor to create boxes quickly.
Select the text area, select the right language, choose an OCR engine and press OCR.
Detect Text Areas and Recognize Text¶
Select the language and an OCR engine, click Edit->Detect text areas and recognize text. Youdou detects in paragragh level and other engines detect text lines. You can use the merge buttons at the right side to merge these areas.
Select the language and OCR engine (only Baidu and Azure support this), click Edit->Detect Balloons.
Heuristic and natural scene text detection methods are also provided, which are more complex. See Text Area Detection
OCR All Text Areas¶
We can mark out text areas first and then OCR them in bulk. Click Edit->OCR All Text Areas to do this.
Sort text areas based on coordinate.
There are several options for exporting.
- Tab-splitted TXT documents include coordinate, font style and text.
- The XLSX file has the same content as TXT’s.
- XLSX file - create worksheets based on folders. It will store picture names, source texts, and target text in separate sheets.
- All text. Store text in separate txt files for each picture.
- Document for translation. Export source text and target text to a docx, txt or XLIFF file.
Enter the target text in the translation area and click Save to complete the translation of a text area.
Translations can be exported as docx documents for external translation, and then import back through File->Import translation.
BasicCAT, a computer-aided translation software, supports direct operation of ImageTrans’s project files for translation.
Translation memory, machine translation and terminology management¶
Switch the tab page on the right to Translation Assistant page to use translation memory, machine translation, and terminology management. You need to set up APIs and enable them to use machine translation. You also need to set the language pair of the project, which can be done through Project->Settings->Select language pair.
Click Project->Batch->Pre-translation to use translation memory or machine translation to pre-translate. Only machine translation is currently supported.
Click Preview translation in the lower left corner to preview the translation. The precision mode will detect text and reconstruct the background. The imprecision mode will just cover the area with the background color.
When Typesetting mode is checked before preview, target text areas will be boxed out. You can readjust their size and location.
Generate the Translated Picture¶
First switch the image scale to 100%, and then click preview to get the translated picture. Click File->Export the current picture to->JPG, and the result will be exported to the out folder in the picture folder. Another option, ORA, supports exporting files to multi-layer image format ORA, which holds layer information for image editing software such as PS, GIMP, and Krita.
In addition, ImageTrans supports exporting as PSDs.
Setting text styles has effects in two processes: Preview translation in ImageTrans and export as PSDs.
Click Project->Settings->Font Style to set the font name, font size, leading, alignment, and so on.
If you want to modify a style, click the style to load the settings and then click Add when the setting is done. Remember to delete the original style. The first style will be the default style.
Because Photoshop requires a special font name, it needs to be obtained from PS. The way is to create a new picture in PS, create a text box, set the font you want, and complete the text editing operation, letting the text box in non-editing status. Then click Read in ImageTrans. Non-Windows systems have to use readFont.jsx scripts.
You can set special font styles for text areas.
Local styles are also supported. Stroke and rotation angles are supported in addition to what the global text style has. When setting the local font style, the interface for the global font style setting will appear. The last new-added style will be saved as the local font style.
Click the Font button on the left to enable the font settings toolbar, which makes it easy to set local styles.
Click the Selection button on the left to enable the multi-selection toolbar. You can adjust the position of multiple text boxes and unify their font styles.
All of the above operations on a single picture can be perforemd in batch through Project->Batch.