Skip to main content

How to extract images from websites

Browse AI can capture images through screenshots or extract image URLs for use in your data workflows. Choose the right method based on whether you need the visual content or the image source.

M
Written by Melissa Shires
Updated today

Understanding image extraction options

Browse AI offers two approaches to working with images:

Method

What you get

Use when

Screenshot

Visual capture (PNG file)

Need visual proof or documentation

Extract image URL

Link to image source

Need to download or reference images

Extract alt text

Image description

Need accessibility text or captions

Extracting image URLs

From a single image

  1. Start training your robot in Robot Studio

  2. Click Capture Text β†’ Just text

  3. Hover over the image

  4. When prompted, choose "Image URL" or "Image source"

  5. Label it (e.g., "Product_Image_URL")

  6. Continue training

From multiple images (galleries)

  1. Click Capture Text β†’ From a list

  2. Hover over the image gallery or grid

  3. Select when the outline covers all images

  4. Browse AI detects:

    • Image URLs

    • Alt text (if available)

    • Any associated captions

  5. Configure how many images to extract

Common galleries:

  • Product image carousels

  • Real estate photo galleries

  • Portfolio showcases

  • Social media image grids

πŸ“– You can use our Google Sheets integration to display the images directly in in Google Sheets.

Taking screenshots of images

When to use screenshots instead

Use screenshots when you need:

  • Visual documentation of how images appeared

  • Proof of content at a specific time

  • Images that might change or disappear

  • Complex visual layouts with text overlay

How to capture image screenshots

  1. During training, click Capture Screenshot

  2. Choose screenshot type:

    • Selection: Draw around specific image

    • Visible part: Current viewport including image

    • Entire page: Full page with all images

  3. Name your screenshot clearly

  4. Continue training

Common image extraction scenarios

E-commerce product images

  • Main product image URL

  • Gallery thumbnail URLs

  • Alt text for SEO insights

  • Image count per product

Real estate listings

  • Property photo URLs

  • Number of photos

  • Floor plan images

  • Virtual tour links

πŸ’‘ Some sites lazy-load images - scroll to trigger loading during training.

Social media images

  • Profile pictures (usually small, consistent URL pattern)

  • Post images (vary in size and number)

  • Story highlights (may require clicks to access)

News and blog images

  • Featured image URL

  • In-article images

  • Image captions

  • Photo credits

πŸ’‘ Extract both URL and alt text for context.

Handling different image formats

Dynamic images

Lazy-loaded images:

  1. Scroll to image location during training

  2. Wait for image to load

  3. Then extract URL

JavaScript-rendered images:

  • Browse AI executes JavaScript

  • Images load normally

  • Extract once visible

Base64 encoded images:

  • Embedded directly in HTML

  • Extract as data URL

  • Larger file size in results

Working with image URLs

What you can do with extracted URLs

βœ… Possible actions:

  • Import to spreadsheet with IMAGE() function

  • Use in automated workflows

  • Pass to image processing tools

  • Create image galleries

  • Download via other tools

❌ Limitations:

  • Browse AI doesn't download image files

  • Can't extract from password-protected images

  • Some sites block external access to images

  • Temporary URLs may expire

Post-extraction processing

In Google Sheets:

=IMAGE(A2)  // Displays image from URL in cell A2

Via Zapier/Make:

  • Pass URL to cloud storage

  • Process with image tools

  • Create backups

Python example:

python

import requests url = "extracted_image_url" response = requests.get(url) # Save or process image

Best practices

During training

βœ… DO:

  • Check if images are fully loaded

  • Test with different image types

  • Label image fields clearly

  • Consider extracting multiple attributes

❌ DON'T:

  • Assume all images load immediately

  • Ignore alt text (valuable for context)

  • Extract unnecessary high-res versions

  • Violate copyright or terms of service

Choosing extraction method

If you need...

Use this method

Image files

Extract URLs β†’ Download separately

Visual proof

Take screenshots

Image metadata

Extract alt text and URLs

Gallery analysis

Capture list of all images

Quick reference

URL extraction only

Troubleshooting image extraction

Common issues

Problem

Cause

Solution

No image URL option

Image is CSS background

Try screenshot instead

Broken image links

Relative URLs extracted

May need base URL

Images won't load

Lazy loading or auth required

Scroll or login first

Wrong image selected

Multiple images overlapping

Use manual selection

URLs expire quickly

Temporary/signed URLs

Extract and process quickly

When images are protected

Some sites prevent image extraction:

  • Right-click disabled

  • Images in canvas elements

  • Watermarked overlays

  • Flash or proprietary viewers

Alternatives:

  • Screenshot the visible image

  • Check for API access

  • Look for public image sources

  • Contact site owner for access

Image extraction limits

What Browse AI cannot do

  • Download and store image files

  • Extract from protected/encrypted images

  • Process or edit images

  • OCR text from images

  • Extract EXIF metadata

Did this answer your question?