Submit your queries and get the image results back as rows - the title, the site it came from, a link to the page hosting it, and two image URLs: the full-size file on the publisher's own server plus a thumbnail. Up to a thousand queries per run, in any language and country. Every row carries its attribution trail, which is what makes this useful rather than just decorative.
Same shape as the other search verticals - a list of queries, a locale, and how many results you want per query.
The default is 100 results per query - worth knowing before you paste a long list, because queries multiply by that number.
Eight columns appear in the export. Six carry data on every row; two are listed last because they have never carried any.
bing.net address rather than the publisher's domain. Fine for quick previews.source and title this is your attribution set, and all three arrive on every row.6 populated columns - plus width and height, which appear but stay empty.
There's no ranking-position column either, so if result order matters to you, preserve the row order of the export.
Of everything in this catalogue, image results need the most care - and, usefully, this is also the tool that hands you the means to be careful.
Google Images surfaces pictures from across the entire web, owned by thousands of different parties. A search result tells you an image exists and where it lives. It confers no right to reuse it, and no amount of tooling changes that.
What makes this dataset genuinely useful is precisely the pair of fields that support doing the right thing: source tells you whose site an image sits on, and link takes you to the page. That's an attribution trail. It's what you need to credit a photographer, request permission from a rights holder, or evidence that someone else is using your imagery.
So: researching what ranks, auditing where your own images appear, and building a reference list are ordinary uses. Republishing what you find, or feeding it into a model you distribute, are decisions with real exposure - and they're yours, not ours. We'd rather write that plainly than let a marketing page imply a permission it can't grant.
Three settings shape a run, and together they decide both what you learn and what you pay.
Breadth - up to 1,000 queries in a single run, each returned row tagged with the query that produced it, so one export can span a whole term list. Depth - results per query, defaulting to 100. That default is the number most worth adjusting: 50 queries at 100 results is 5,000 rows. Locale - language and country are separate settings across all languages and countries, exactly as on the Google Search Scraper, because what ranks visually differs sharply between markets.
A few examples of how teams use image results.
Search your product names and campaign terms, then read source and link. Sites reusing your photography without permission surface as rows with someone else's domain attached - and you have the page URL to act on.
Image results are a market's visual language. Pulling them for a category shows which aesthetic dominates and which sites own it - useful before commissioning photography that will sit next to all of it.
Run the same query with language and country varied. What ranks for a product term in Japan often looks nothing like what ranks in Germany, and that gap is a concrete input for localisation work.
No subscription, no minimum, no per-query licence. Your first 500 rows are on us - after that, pay-as-you-go.
Every new account, one-time. No credit card required. All scrapers unlocked, full feature set - enough to test your queries and see what a row actually contains.
Roughly $2 per 1,000 results. Read the estimator on this one: at the default 100 results per query, a modest term list becomes a large run faster than on most tools.
Volume pricing, SLAs, dedicated workers and bespoke onboarding for recurring visual monitoring at scale. Tell us your numbers and we'll quote.
LIVESCRAPER10 at checkout.Two tools share this one's discipline: you get references and attribution, never a licence.
Photo links from a place's Maps listing rather than from a web search. The right tool when the question is what a specific location looks like.
The organic vertical for the same query set - listings, ads, shopping results and related questions. Run both to see who owns a term in text as well as in pictures.
Short answer: yes for the results page. The images themselves are a separate question, and an important one.
An image results page is public. Google shows anyone the titles, sources and links for a query, signed in or not. Collecting publicly visible results for research and monitoring is long-established practice, and as long as the data is publicly available and the process doesn't disrupt the site there are no federal laws prohibiting it.
The images are different. They belong to whoever created them - photographers, brands, publishers, individuals - and appearing in a search result changes nothing about that. This tool is built for discovery and attribution: it tells you what exists and whose it is. Reuse is a conversation between you and the rights holder, which is exactly why link is in the output.
Google's terms also restrict automated access, so this is a terms question as well. We touch nothing behind a login, run no third-party trackers on the data layer, and your exports auto-delete after 30 days.
The questions we hear most. Anything else? Talk to us - humans, not bots, write the answers.
query, title, image, thumbnail, source and link - the query that surfaced it, the image's title, the full-size file on the publisher's own server, a small preview served from a third-party CDN, the site it belongs to, and a link to the hosting page. A width and height column also appear but we have never seen either carry a value. There's no ranking-position column, so preserve the export's row order if result order matters.image points at the full-size file on the publisher's own server and thumbnail at a small preview on a third-party CDN. Both arrive on every row, as do the attribution fields source, link and title. Fetching or storing anything from those URLs is a step you take, which also keeps the licensing question in your hands - see the note above.link is in the output precisely so you can find the owner and ask.source is a domain that isn't yours. You get the hosting page via link, which is what you'd need to raise it. Bear in mind this is keyword-driven discovery rather than a reverse-image match, so treat it as a broad sweep.500 one-time free rows on every new account - no expiry. After that it's $0.002 per row, pay-as-you-go - no card on file until you say so.
Livescraper's Google Search Images Scraper turns a list of queries into image result data. You paste the terms you want covered - up to a thousand per run - set the language and country, choose how many results to return per query, and download the output as CSV, JSON or Excel.
Each row carries six populated fields: the query that surfaced the result, the image title, the source site, a link to the page hosting it, and two image URLs - the full-size file on the publisher's own server and a smaller preview served from a third-party CDN. A width and height column also appear in the export but have never carried a value in our own runs, so treat them as present but unpopulated. The pair worth focusing on is source and link - together they form an attribution trail, telling you not just that an image exists but whose it is and where to find the owner.
That attribution is what makes the dataset practical. Brand and legal teams search their own product terms and look for rows whose source domain isn't theirs, which surfaces unauthorised use along with the page URL needed to raise it. Creative teams read a category's visual language before commissioning photography that will sit alongside it. International teams vary language and country on the same query, because what ranks visually in one market is often unrecognisable in another.
One thing to be clear about: a search result is a reference, never a permission. Images belong to whoever created them, and appearing in a result changes nothing about that - which is precisely why the hosting link is in the output. Note also that the default is 100 results per query, so fifty terms becomes five thousand rows; adjust depth before a broad sweep. Start free: your first 500 rows cost nothing and need no credit card.