Reverse image search and visual product discovery

Reverse image search with Google Lens

Reverse image search helps you discover what appears in a picture, where the picture is used online, and which products look like it. Add one or more image links or upload a file, choose what you want to find, and start the run. The results are organized into clear tables that you can review or download.

Use this Actor when you want to:

• find pages that use the same or a similar image;
• identify products from a photo;
• compare sellers, prices, ratings, and availability;
• find the original source of an image;
• extract or translate text from a picture;
• ask questions about an image;
• get help with a photographed homework problem.

How to use the Google Lens search

1. Open the Input tab.
2. Paste one or more public image links into Image URLs, or open Upload and choose a file from your device.
3. Keep the default Visual matches and OCR, or choose only the result types you need. Select All modes only for broad experimental coverage.
4. Select the language and country that best match the results you want.
5. Click Start.

When the run finishes, open the Dataset tab. Separate views help you move between visual matches, products, exact matches, text, answers, and related search ideas.

Choose what you want to find

Search mode	What it gives you
All modes	Runs every available type of image analysis
Visual matches	Finds native Google Lens visual cards and the pages where they appear
Products	Finds matching products, sellers, prices, ratings, and availability
Exact matches	Finds pages using the same image and can look for the original file
OCR	Extracts text from an image
Translate OCR	Extracts the text and translates it into your chosen language
AI Mode	Answers questions about what is shown in the image
Homework	Reads a photographed problem and returns an answer with steps when available

What data can you get?

Every image receives an overall status and its own group of results.

Result	Included information
Native visual matches	Page title, website, page link, preview image when Google supplies one, and short description
Related web results	Clearly labeled semantic fallback links returned when native visual cards are unavailable
Product matches	Product name, seller, price, currency, availability, rating, reviews, and image
Exact matches	Matching page, publisher, publication date, preview, and original image link when found
Related results	Suggested searches, related pages, and other useful image search directions
Image text	Full text plus words, lines, and paragraphs
Translated text	Extracted text in the selected target language
AI answers	Your question, the answer, and supporting links when available
Homework	The recognized problem, answer, and solution steps when available

Download the Dataset as Excel, CSV, or JSON, or schedule runs in Apify.

Popular use cases

Find similar images online

Upload a photo to find visually similar images, their source websites, preview pictures, and related searches. This is useful for creative research, content discovery, catalog work, and checking where an image appears online.

Find a product by image

Use a product photo, screenshot, or catalog image to find possible matches. Product results can include seller links, prices, currencies, availability, ratings, review counts, and product pictures. Choose the country that matches the market you want to search.

Find the original image source

Select Exact matches and turn on Resolve original image URLs. The results can include the matching page, publisher, publication date, original image link, and a confidence value. This makes it easier to trace an image back to an earlier or higher-quality source.

Extract text from an image

Select OCR to turn text in photos, screenshots, labels, documents, and product packaging into text you can copy or download. Choose full details, paragraphs, lines, words, or text only. Select Translate OCR when you also want the text translated.

Input options

Option	What to enter
Image URLs	One or more public links to JPG, PNG, WebP, GIF, BMP, or TIFF images
Upload an image	One image file from your device
Search modes	One or more result types, or All modes
Results per search mode	Target/cap for saved matches; Google still controls availability
Visual search depth	Standard follows native continuation pages and the Visual Matches tab; Extended also searches salient image regions
Maximum native visual queries	Full-image plus crop upload limit for Extended mode

Native Visual Matches use last-known-good session semantics. A challenged or failed request cannot update stored cookies, and the scraper can retry once on a new anonymous sticky residential identity. Separately, a scheduled headed CloakBrowser maintenance Actor validates Chrome DevTools-staged cookies and promotes them only after main=200 and native folsrch=200; see docs/SESSION_REFRESHER.md. | OCR detail | Full details, paragraphs, lines, words, or text only | | Translate OCR to | The language for translated image text | | AI Mode questions | Up to 10 questions about each image | | Result language | The language used for recognition and search results | | Country | The market used for localized product and search results | | Resolve original image URLs | Whether to look for the original file on matching pages |

Example

Imagine you have a photo of a red running shoe and want to identify it.

1. Add the image link or upload the photo.
2. Select Products, Visual matches, and Exact matches.
3. Choose your country and start the run.

The Dataset may contain similar shoes, possible product names, merchant pages, prices, ratings, and pages that use the same photo. Switch between the Products, Visual matches, and Exact matches views to review them.

Visual match counts and pricing

maxResults is a target and output cap, not a promise. In current first-page native Lens responses, Visual Matches commonly contains about 3–6 cards per image. Standard mode follows fresh native folsrch continuation links when Google supplies them, then checks the native udm=44 Visual Matches tab without another upload. Extended mode then searches up to five additional salient regions, deduplicates canonical source URLs, and stops at the target or after two crop queries add no unique results.

Every native result identifies whether it came from the full image or a crop, the crop coordinates, query number, native page, native rank, and every query that found the same URL. Continuation URLs are short-lived and are never included in output.

The current price is shown in Apify before a run starts. The Actor records which successfully delivered features are billing-eligible. A native visual event is counted for each full-image or crop query that adds at least one unique result. Continuation pages do not add another native-query event. Semantic fallback links are never labeled or billed as native visual matches. Failed images do not produce success events.

For the exact pipeline, field definitions, stopping rules, operating metrics, and rollout procedure, see docs/EXTENDED_VISUAL_SEARCH.md.

Frequently asked questions

How can I reverse Google search an image?

Add a public image link or upload a file, then select Visual matches and Exact matches. Start the run and open the Dataset to review similar pictures, matching pages, source websites, and original image links when they can be found.

How do I find an image that is similar to a picture?

Choose Visual matches for the image you want to check. Standard mode uses the full image, native continuation pages, and the dedicated Visual Matches tab. Choose Extended to search additional image regions when a larger result set is worth the extra requests. The Actor cannot guarantee the requested count.

Can I find a product from a picture?

Yes. Select Products and choose the country where you want to shop. The Actor can return possible product names, seller pages, prices, currencies, availability, ratings, review counts, and product images.

How can I extract text from an image?

Select OCR, add the image, and start the run. The Dataset can contain the full text as well as separate words, lines, and paragraphs. Choose Translate OCR if you also want the extracted text translated.

Is this an official Google Lens API?

No. This is an independent, unofficial scraper and is not affiliated with, endorsed by, or operated by Google. Google can change the underlying response formats and result availability. Use it in accordance with the terms and laws that apply to your use case.

Can I use it through an API?

Yes. The same Actor can be run from Apify or through the Apify API. Review the current price displayed by Apify before starting a large batch.