AI Background Remover – Transparent PNG & Product Photos in Bulk

Remove the background from hundreds of images at once with an AI segmentation model (IS-Net). Give it image URLs, a CSV or a dataset from any scraper. You get transparent PNG/WebP cut-outs, or product photos on white or any colour, plus masks and auto-crop. Every result has a direct download link.

• E-commerce ready: white background for Amazon, eBay, Etsy and Shopify product images, or keep it transparent.
• Soft or crisp edges: natural edges for hair and fur, or a hard threshold for products, logos and objects. Optional feathering.
• Auto-crop to subject with a margin, or mask only output for your own compositing.
• Bulk & automation: lists, Google Sheets/CSV, datasetId from another Actor, API, schedules, webhooks, Make/Zapier/n8n.
• Pay per image, no GPU, no subscription. Failed images are free.

Pricing

$0.005 per image (background-removed). Errors, invalid URLs and skipped images are free. The run stops cleanly at your maximum cost per run.

Input example

{
  "imageUrls": ["https://example.com/product.jpg"],
  "background": "white",
  "outputFormat": "jpg",
  "cropToSubject": true,
  "cropMargin": 40,
  "edgeThreshold": 128
}

Field	Default	Notes
background	transparent	transparent, white, black, color (with backgroundColor)
outputFormat	png	png / webp keep transparency, jpg uses a solid background
cropToSubject, cropMargin	off, 0	Trim the empty space around the subject
outputMaskOnly	off	Black/white alpha mask
edgeThreshold	0	0 = soft AI edges, ~128 = crisp edges
featherPx	0	Blur the edge for a softer cut
maxInputMegapixels	25	Bigger inputs are skipped for free

Output

{
  "url": "https://example.com/product.jpg",
  "status": "ok",
  "outputUrl": "https://api.apify.com/v2/key-value-stores/<id>/records/nobg-0001-product-1a2b3c.png",
  "inputWidth": 1200, "inputHeight": 1200, "outputWidth": 980, "outputHeight": 1104,
  "format": "png", "background": "transparent",
  "subjectCoveragePercent": 41.7, "outputBytes": 845123, "processingSeconds": 4.2
}

subjectCoveragePercent is the share of the image detected as foreground. If it is below 0.5%, the row gets a warning; landscapes or scenes without a clear subject can't be cut out. Error rows are free: HTTP_404, TIMEOUT, TOO_LARGE, INPUT_TOO_LARGE, NOT_AN_IMAGE, DECODE_ERROR.

Use cases

• Product photos on white or transparent backgrounds for marketplaces and stores
• Profile pictures, team photos and ID-style photos on a solid colour
• Stickers, print-on-demand designs, thumbnails, ads and social posts
• Car, furniture and real estate listings
• Catalogue pipelines: scrape product images, then cut them out here

Notes

• The model works best on images with a clear main subject (people, products, animals, objects). Very fine detail (individual hairs, glass) may be softened.
• Only process images you have the right to use.

Credits

Segmentation model: IS-Net / DIS "general use" (Apache-2.0), ONNX build distributed with rembg (MIT), run with ONNX Runtime. Not affiliated with the model authors.

Related Actors by gazidev

AI Image Upscaler · Bulk Image Converter & Compressor · Image Downloader · Image Metadata & EXIF Extractor