Google Lens Scraper

🔍 Google Lens Scraper

Reverse image search at scale. Give it image URLs or base64 images and get back every page Google Lens matches: page title, link, site, thumbnail and the full-size image URL. No Google account, no API key, no browser setup. Results land in a clean table you can export to JSON, CSV or Excel.

✨ Why use Google Lens Scraper?

• 🖼️ Three Lens tabs in one run: visual matches, pages that contain the image, and exact matches with publication date and image size.
• 🔗 Real page links: Google's redirect links are resolved to the destination page, and every match carries its full-size image URL when Lens shows one.
• ⚡ Fast: one image with visual matches and pages-with-this-image takes about 35 seconds end to end (median of our test runs). Several images run in parallel.
• 💸 Cheap: you pay $0.80 per 1,000 matches plus a few cents of platform usage per image. No start fee.
• 🧾 Clean output: one row per match, same columns for every match type, ready for spreadsheets and databases.
• 🛡️ Robust: blocked lookups are retried on fresh browser sessions, and an image that still fails gets a clear status row instead of breaking the run.

🎯 Use cases

• 📸 Find where a photo is published: track reuse of your product shots, artwork or press photos across the web.
• 🛒 Product research: find shops and marketplaces that show the same or similar product, with price text when Lens shows it.
• 🕵️ Verify images: check the earliest dated copies of a picture for fact-checking and trust and safety work.
• 🧑‍🎨 Copyright and licensing: collect the pages that use an image so you can follow up with the site owners.
• 🔎 SEO and brand monitoring: see which sites rank in Lens for your images and logos.
• 🧠 Dataset building: gather visually similar images and their source pages for research.

🚀 How to use

1. Paste one or more public image URLs into Image URLs, or paste base64 images into Base64 images.
2. Set Max results per match type (default 20).
3. Pick the Match types you want: visual matches, all (pages with this image) or exact matches.
4. Click Start and wait for the run to finish.
5. Open the Output tab and export the table to JSON, CSV, Excel or HTML, or read it through the API.

📥 Input

{
  "imageUrls": [
    { "url": "https://upload.wikimedia.org/wikipedia/commons/a/a3/Eq_it-na_pizza-margherita_sep2005_sml.jpg" }
  ],
  "searchTypes": ["visual-match", "all"],
  "maxResults": 20,
  "language": "en"
}

📤 Output

One row per match. Images that give no matches get one status row with the error column filled.

Example rows from a real run:

[
  {
    "searchType": "visual-match",
    "position": 8,
    "title": "Marinara sauce - Mayo Clinic Press",
    "link": "https://mcpress.mayoclinic.org/health-letter/marinara-sauce/",
    "source": "Mayo Clinic",
    "domain": "mcpress.mayoclinic.org",
    "thumbnail": "https://encrypted-tbn2.gstatic.com/images?q=tbn:ANd9GcTEa3uwXXGIU5Koezl7eLEzuJh-D25Pkc-qBDiabeyVMKEBNxJT",
    "imageUrl": "https://mcpress.mayoclinic.org/uploads/2026/03/MarinaraSaucexGettyImages-707588669-1120x640.jpg",
    "imageWidth": 1120,
    "imageHeight": 640,
    "date": null,
    "price": null,
    "inputUrl": "https://upload.wikimedia.org/wikipedia/commons/a/a3/Eq_it-na_pizza-margherita_sep2005_sml.jpg"
  },
  {
    "searchType": "exact-match",
    "position": 1,
    "title": "Learn How To Make A Perfect Margherita Pizza - Italy Villas",
    "link": "https://www.italy-villas.com/to-italy/2015/food-and-wine/recipes/pizza-margherita-recipe",
    "source": "Italy Villas",
    "domain": "italy-villas.com",
    "thumbnail": "https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcTDFBtOB0m_ufgeBENxH08FiMUb3Xr8PDXWQbxbxANVQ2--LBTHXSj6l9Gd&s",
    "imageUrl": null,
    "imageWidth": 1024,
    "imageHeight": 768,
    "date": "Sep 23, 2015",
    "price": null,
    "inputUrl": "https://upload.wikimedia.org/wikipedia/commons/a/a3/Eq_it-na_pizza-margherita_sep2005_sml.jpg"
  }
]

💰 Pricing

• $0.80 per 1,000 matches ($0.0008 per saved row). Images that fail are not saved and never charged; they are listed in the run's status message and in the OUTPUT record (failedInputs). No start fee.
• Platform usage (hosted browser time, a little compute and datacenter proxy traffic) is billed on top. In our test runs it was about $0.04 per image.
• Worked example (measured): 1 image with visual matches and pages-with-this-image returned 117 matches for $0.094 in result fees plus $0.041 of platform usage, about $0.13 in total.
• At scale: 69 images with all three match types returned about 10,000 matches in roughly 25 minutes, about $8 in result fees plus about $2 of platform usage.
• Set a maximum total charge on the run to cap spending. Platform usage counts toward that cap too, so allow at least $0.10 per image; the run stops when the cap is reached and keeps everything saved so far.

❓ FAQ

Is it legal to scrape Google Lens? It reads public search results only, the same pages anyone can open without logging in. You are responsible for how you use the data, especially personal data and copyrighted images.

Do I need a Google account, cookies or my own proxy? No. The lookup runs in a hosted browser with its own network. The proxy setting only affects downloading your image URLs and resolving exact-match links.

How many results can I get per image? As many as Lens shows: usually 40 to 60 visual matches, a similar number of pages with the image, and up to about 150 exact matches. Set Max results per match type higher to keep all of them.

Why are some fields empty? Lens does not show every detail for every match. Full image URLs exist for visual matches and pages-with-this-image; dates, image sizes and prices appear mostly on exact matches.

What happens when Google blocks a lookup? The image is retried on fresh browser sessions. If Lens still refuses, the image gets a status row and the run moves on to the next image.

Can I schedule it or call it from code? Yes. Use the scheduler for recurring runs, or start runs and read the dataset through the API and client libraries.

⚠️ Limits

• Images over 8 MB are skipped with a status row.
• The whole run input must currently stay under about 8 KB. That fits about 50 pasted image URLs; for more, link a text file with one URL per line. Base64 images only fit when they are tiny (a few KB), so prefer image URLs.
• Lens AI answers, text recognition (OCR), translation, homework help and product shopping cards are not extracted.
• Thumbnails are Lens preview URLs, not embedded image bytes.
• An exact-match link that cannot be resolved stays as Google's redirect link, which still opens the right page.
• Coming from another Lens actor? Visual-match rows keep position, title, link, source and thumbnail; web results use link instead of href, and every match is its own row instead of a nested list.