Yandex Reverse Image Search 🔍 File Upload, OCR & All Regions
Pricing
Pay per usage
Find where any image appears online. Search by picture instead of keywords: paste a link or upload a file and get matching pages, visually similar images, other resolutions, text read from the picture, and descriptive tags.
The web address of the image you want to search for. Right-click any picture in your browser and choose 'Copy image address', then paste it here. The link must be publicly reachable — a private or password-protected link will not work. Leave this empty if you are uploading a file below.
Upload a picture from your computer instead of pasting a link. Accepts JPG, PNG, WebP, GIF and BMP, up to about 10 MB. Use this for screenshots, photos, or any image that is not already published online.
Optional. Search a rectangle inside the picture instead of the whole thing — useful when one object in a busy photo is what you care about. Four numbers between 0 and 1 separated by semicolons: left;top;right;bottom. For example 0.25;0.25;0.75;0.75 searches the middle. Leave empty to search the whole image.
Which regional Yandex indexes to search. Leave as-is for most uses. Adding Turkey is the single best way to find more results, because it searches a largely different index — expect roughly 75% more unique matches. The Russia, Belarus, Kazakhstan and Uzbekistan indexes overlap almost entirely with the International one, so adding them returns only a handful of extra results. Each extra region makes the run take a little longer. Picking 'All regions' searches every index and overrides the other choices.
[ "yandex.com"]What kind of results you want back. Every result is one row in the dataset, labelled with a 'resultType' column so you can filter them. 'Pages containing this image' and 'Visually similar images' are on by default and cover most needs.
[ "matchingPages", "similarImages"]The largest number of rows to take from EACH search region, not from the run as a whole. With 3 regions selected and a limit of 11, the run takes up to 11 from each, so up to 33 rows in total. Duplicate removal then runs across regions, so the final count is usually lower — the same picture found in two regions becomes one row. Set to 0 for everything found: a single region can return 3,000+ rows and takes correspondingly longer. Within each region's own limit, the rows are shared fairly between the result types you selected.
On (default): if the same picture is found on several pages, you get one row listing every page that carries it, in a 'pages' column. This removes roughly 40% of rows without losing any information. Off: one row per page, with the same image repeated many times.