If you can select the text, right-click it and choose Translate, paste it into Apple's Translate app, or use Safari's webpage translation. For text that cannot be selected, such as a scan, subtitle, image or app interface, capture the relevant screen region and run OCR before translation. If image text should be selectable but is not, use the Live Text troubleshooting guide.
Translate text in images, PDFs and videos
OCR can read visible text from a scanned PDF, screenshot, burned-in video subtitle, game or untranslated app interface. This avoids retyping the words and works even when the source app offers no text selection.
Capture and translate a screen region
Screen Translate works on the screen itself: press the hotkey (Option-Shift-T by default), drag over the region, and it recognizes the text, detects the language, and shows the translation side by side with what it read. If the OCR grabbed a word wrong, edit the captured text right there and translate again - without leaving the window.
Recognition uses Apple's Vision framework and translation uses Apple's on-device Translation framework. After macOS downloads the required language resources, the process works without an internet connection. Supported languages depend on the macOS version installed on your Mac.
Free, open source, and downloadable from GitHub.
FAQ
How do I translate text in an image or PDF on a Mac?
If Live Text can select it, copy it into the Translate app. If it can't - scans, videos, app interfaces - a screen translator OCRs the region you select and translates it in one step.
Does it need an internet connection?
No. Recognition and translation both run on-device. The first time you use a new language pair, macOS downloads a language pack; after that it's fully offline.
What languages are supported?
The languages of the macOS Translation framework - currently 22, including Chinese, Japanese, Korean, Arabic, Ukrainian, and the major European languages. Availability can vary by macOS version.