- Auditing a supplier image batch to see which photos carry watermarks or text before relisting.
- Deciding scope for Smart Erase — knowing whether a mark is on the background or the product itself.
- QA before upload: catching a stray logo or QR code that would get a listing rejected.
// image · low cost
Element Detection Spot text, logos, and watermarks in your product images.
// Upload image → detect text / logo / watermark / text-color-block positions
Returns position + type as JSON
Returns position + type as JSON
Before you clean up a batch of product photos, you need to know what is actually on them — which images carry watermarks, embedded text, logos, or QR codes, and where those marks sit. Element Detection scans an image and reports the elements it finds, so you can route each photo to the right cleanup step instead of running an expensive erase pass on images that do not need one.
It is one of the cheapest tools in the catalog because it only inspects — it never modifies your image. Use it as the front door to the photo workflow: detect first, then send the flagged images to Smart Erase or Background Cutout. On large catalogs this saves credits and avoids erasing things you meant to keep.
// Perfect for
- Amazon FBA sellers pre-scanning wholesaler photo batches for compliance risk
- Shopify dropshippers auditing AliExpress supplier images before storefront upload
- TikTok Shop sellers flagging watermarked photos before mass upload
- Compliance teams running due-diligence on bulk supplier image drops
- Input: JPG / PNG / WebP, ≤10 MB
- Output: detection report (element type + location), image unchanged
- Price: ≈ $0.0078 / image (≈ 0.78 credits)
- Trial available · failed jobs auto-refund
Frequently asked questions
Does Element Detection change my image?
No. It only inspects and reports what it finds — text, watermarks, logos, QR codes — and returns your original image untouched. Removal is a separate step in Smart Erase.
Why run detection before erasing?
It tells you where a mark sits so you can scope the erase correctly (background-only vs subject-only), which gives cleaner results and avoids erasing detail you wanted to keep. On big batches it also lets you skip images that need no work.
Can it detect text in any language?
It detects the presence and location of text regions regardless of language. If you then need to translate that in-image text rather than erase it, use Image Translation instead.
What happens to my images afterward?
Auto-deleted from processing servers within 24 hours, encrypted in transit, and never used to train any AI model.