Translating the text of images: zoom in on this intelligent translation technology

Intelligent translation – translating text from photos – is a rapidly developing technology. Software and applications for translating text from images allow users to obtain near-instantaneous automatic translation from all types of visual material. Here's a closer look at how this type of translation works; its advantages and limitations compared with human translation.

What is an image translator?

New technologies based on artificial intelligence (AI) continue to evolve exponentially, and image recognition is no exception. This AI technology is used in particular to automatically translate text from photos, so that its content can be understood instantly.

This method can make everyday life easier for foreign speakers, who can now translate a sign, a menu, a label, or a document simply by taking a photo of it with their mobile phone or computer.

How does photo translation work?

Instant translation of text in photo format is made possible by a mobile app, software, online platform, or smartphone with built-in translation functionality, which can be based on a range of AI technologies, including:

Translation based on optical character recognition (OCR)

This technology extracts text from an image, analyses it and automatically translates it into the target language. With annual growth of 13% (Source: exactitudeconsultancy.com), the OCR market meets the needs of a multitude of sectors and industries across the globe.

Many image translation applications and software packages are now based on this technology, including:

• Google Translate: this free tool, used by almost 500 million users worldwide (Source: phonandroid.com), lets you automatically translate an image using your phone's camera, or by uploading the image to the translation application.
• Microsoft Translator: this software also has a text-to-image translation function.
• Text Fairy: an image translation and scanning application for Android.
• Scan & Translate +: iOS application used to scan and translate texts in photos.
• Prizmo: OCR software for MacOS with scanning and image translation functions.

Translation based on image segmentation

Some photo translators divide images into several segments to analyse all their parts. This step may require advanced image processing techniques (such as character recognition or object detection), to easily locate text on a complex visual. Translation based on image segmentation is particularly suitable for computer graphics or diagrams. It is used by a number of intelligent tools, such as:

• Google Translate: uses image segmentation to detect and isolate the text in a photo. It then applies OCR to convert the photo, before translating it into the target language.
• Microsoft Translator: also uses this technology to extract the text from your photo, before translating it into the language of your choice.
• The Waygo application specialises in translating text from photos in Asian languages such as Chinese, Japanese, and Korean, using image segmentation.

Translation based on contextual localisation 

Contextual localisation refers not only to the analysis of text in images, but also to the visual context in which it is embedded (objects, icons, logos, etc.). Taking the latter into account enables the automatic translator to be more precise by adapting to the target market. This is particularly useful in the marketing, advertising, and e-commerce sectors.

Translation tools based on contextual localisation include DeepL (online machine translation system) and Mate Translate (browser extension).

Computer Aided Translation (CAT)

Computer-assisted machine translation (CAT) makes use of translations that have already been produced by organising them and storing them in a dedicated database. This technology is used in a variety of ways, from standalone software solutions to mobile applications. It is often used to translate social media content, e-commerce sites, or documents. CAT applications and software include:

• SDL Trados Studio
• Déjà Vu
• Smartcat

Automatic photo translation VS professional translator: which should you choose?

Automatic translation of text into photos is simple, easy to access, and often free of charge, but it does not always produce high-quality results. Results vary depending on the quality of the photo (pixelation, clarity, contrast, etc.), and errors in image recognition and text extraction can sometimes occur. You should therefore consider using the services of a professional translator to guarantee a reliable result.

Alphatrad has a network of 3,500 specialised professionals including highly qualified native translators. These translators translate into their mother tongue and can produce multilingual translations of your content in image format. These translators specialise in your business sector, they work in over a hundred languages, and can translate photos into English, Italian, French, Arabic, German, Spanish, etc.

Our teams of translators can translate all types of documents in a wide variety of formats, including images of all kinds:

• Restaurant menus
• Computer graphics
• Photos of printed documents (books, manuals, tourist guides, etc.)
• Photos of handwritten documents (medical notes, reports, summaries, etc.)
• Visuals for marketing purposes (advertising posters, flyers, brochures, etc.)
• Product labels and packaging
• Scanned documents (notarial deeds, supporting documents, passports, identity cards, contracts, etc.).

FAQ on photo translation