WIPO's free Global Brand Database uses deep neural networks to match trademark images by concept — apple, eagle, star — rather than keywords, pooling close to 38 million marks from 45 offices' national collections since its April 2019 launch.
45
Participating trademark offices (2019)
~38,000,000
Registered marks covered at launch (2019)
2019
Launch year
Details
Maturity
Established
Promoter
World Intellectual Property Organization (WIPO)
Period
2019-present
Keywords
intellectual property, trademark search, government digitalisation, AI-assisted search
Context
Launched in April 2019 and integrated into WIPO's Global Brand Database, this free tool uses deep machine learning, trained on figurative-element classification data from the Madrid System and major trademark offices, to recognise visual concepts within a trademark image — an apple, an eagle, a crown, a star — rather than relying on keyword tags.
Results
The tool covers the national trademark collections of 45 participating offices, representing close to 38 million registered marks at launch, including offices that had never used a figurative-element classification system before. WIPO states the tool 'delivers results in a second' and produces 'labor-cost savings for trademark examiners, attorneys and paralegals', but this is WIPO's own characterisation, repeated by independent IP trade press, rather than a quantified or externally audited evaluation.
Implementation
Indicative cost
Medium (€50k–€500k)
Time to results
Long (> 3 years)
Staffing & skills
WIPO Global Brand Database technical team (Madrid System)
Conditions for success
participation and data-sharing from national/regional trademark offices
figurative-element classification training data from the Madrid System
Common failure modes
no independently audited accuracy or time-saved evaluation has been published
Commonly funded by
Own resources / municipal budget
Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.
Do you run this practice?
Claim it —
verified implementers get a public contact pathway and can propose corrections.
Data sources
Where this practice's information was retrieved from, and when.
Norway's national open-data catalogue, data.norge.no (run by Digdir), added an AI-powered natural-language search built on Google Vertex AI, letting users …
