Contract: Image-text training dataset for multimodal image translation
This contract covers image-text training dataset for multimodal image translation for a global machine translation company.
- Contract date
- Service area
- Data Development & Processing
- Client type
- A global machine translation company
- Role
- Contractor
Contract overview
This contract covers image-text training dataset for multimodal image translation for a global machine translation company.
Scope of the contract
- Collected and cleaned images including signs, menus, documents and product information.
- Extracted in-image text with OCR.
- Built Korean, English and Japanese translation data and verified image-text alignment.
- Generated metadata and structured a large volume of items as JSON/CSV training data.
Technologies & capabilities
Multimodal data · OCR · Multilingual translation data · Data structuring
Contract scope at a glance
