Google Images Scraper

Collect bounded, public Google Images results with exact full-image URLs, Google thumbnail URLs, source-page URLs, titles, domains, and available dimensions.

Behavior

• Returns at most 100 unique image results for one query, subject to available results.
• Supports country, interface language, SafeSearch, size, type, and color filters.
• Uses the Google SERP proxy by default. Residential, custom and direct browser routes are optional; direct and US Residential requests were rate-limited in the latest checks.
• Select the GOOGLE_SERP Apify proxy group to fetch Google Images HTML through the SERP route. This route uses HTTP without proxy sessions and extracts matching image/source metadata from the same result ID. It never executes returned scripts or downloads images.
• Loads thumbnails with bounded scrolling, then opens results and reads each selected card's own Google image link to verify its full-image URL, source URL and dimensions. A thumbnail is never substituted for the full image, and unrelated or stale panels are never paired with a new result.
• Emits the historical stable recordId, recordType, fieldCoverage, and dataQuality fields alongside source-backed business fields, omitting null or empty values.
• Stores run diagnostics in OUTPUT; raw Google HTML and failure rows are never written to storage.
• Fails closed on rate limits, challenges, consent pages, malformed source state, or image/source pairing failures.
• Uses bounded retries for transient failures and redacts credentials and query-string secrets from failure evidence.

Input example

{
  "query": "red fox wildlife",
  "maxItems": 5,
  "gl": "us",
  "hl": "en",
  "imageSize": "large",
  "imageType": "photo",
  "imageColor": "any",
  "safeSearch": "active",
  "proxyConfiguration": {
    "useApifyProxy": true,
    "apifyProxyGroups": ["GOOGLE_SERP"]
  }
}

The Input tab documents all supported values and the bounded timeout, delay, and retry controls.

Dataset

Every dataset row is an image-result with a stable recordId, verified imageUrl, thumbnailUrl, and sourceUrl values plus position, title/alt text, source/domain, search context, provenance timestamp, optional dimensions, fieldCoverage, and dataQuality. Operational diagnostics remain in OUTPUT.

When Google explicitly reports no image results, the Actor completes as a valid empty run. Access barriers and structurally ambiguous pages fail the run and leave the dataset empty.

Running and downloading results

Enter your search in the Actor's Input tab and start the run. After a successful run, open the Dataset tab to review the records. You can download the dataset in various formats such as JSON, HTML, CSV, or Excel. Set proxyConfiguration.useApifyProxy to false for an explicit direct test. The default requests the Google SERP proxy; a failed proxy setup never silently falls back to direct access.

Developer Options

requestTimeoutSecs bounds navigation to 15–90 seconds (default 60). maxRetries allows 0–2 fresh attempts after a transient failure (default 1); explicit access restrictions stop immediately. minDelayMs and maxDelayMs each accept 0–5,000 milliseconds (defaults 400 and 900), with minimum no greater than maximum. Image clicks wait at least one second for the selected result to populate its source link.

Costs and limitations

Browser time and Residential proxy traffic incur Apify usage. SERP proxy usage is charged per request. Larger limits require more image clicks on the browser route; the SERP route is bounded by the results in one HTML response. Google can rate-limit or deny requests even with a proxy; access restrictions are reported as failures rather than successful empty datasets. Optional snippet, publisher, favicon and licensing values are omitted when the matched source does not expose them. Images are referenced by URL, not downloaded; users must obtain any necessary rights from the source publisher.

Google Images markup and anti-automation controls change frequently. A blocked run is intentionally visible in OUTPUT.failureEvidence; the Actor never turns a challenge page into guessed image records.