IMAGE
COMMERCIAL
RAW
On the SEA-Vision benchmark (arXiv 2603.15409, March 2026), Qwen3-VL-32B reads English scene text at 72 percent, Filipino at 24, Thai at 28, Vietnamese at 29, and Khmer, Burmese and Lao lower still. Training data for those languages barely exists: MTVQA covers only Thai and Vietnamese from the region, and commercial OCR catalogs skip the rest. Contributors in 11 Southeast Asian countries, most of them in Indonesia, Malaysia and the Philippines, photograph packaging whose main text is not English, holding the phone square to the panel so the type stays sharp; English-only packs and unreadable text are refused. There are no transcriptions, and the panel-first framing makes line-level polygons quick to add.
Volume
50,000+
Photos
Contributors
2,000+
Regions
Southeast Asia (11 countries)
Updated
What’s in the collection and how it’s delivered.
Data modality
Images
Annotated
Raw (no labels)
Exclusivity
Non-exclusive rights
Capture tool
Phone camera (in app)
Use cases
Multilingual scene-text recognition and OCR, text-centric VQA in Southeast Asian languages, product matching
Format
JPG / PNG / HEIC
Resolution
Native phone camera resolution (JPG, PNG, HEIC)
Total volume
50,000+ photos (approx. 150 GB)
Regions
Southeast Asia (11 countries)
Purity level
99% passed QA review
Where this data comes from and how it’s licensed.
Every file comes straight from the contributor who provided it and is rights-cleared before it’s listed, so your license traces back to the source.
Contributor
Provides the data and uploads it straight to Kled.
Kled
Clears the rights before anything is listed.
Your team
Licenses it under the terms below.
License terms
Non-exclusive commercial training license.
Consent
Each contributor opted in to this use when they uploaded.
Specs reflect the current collection and may change as new data is added. Final terms are set in your license agreement.
