Dance Recognition
A System for Indexing, Monitoring, and Protecting Dance IP
The world of dance is a vibrant expression of artistry, yet its creators often lack the tools to protect their intellectual property. In the digital era, platforms like TikTok have revolutionized the dissemination of dance, yet creators of dance trend face significant challenges in safegaurding their moral right of attribution. As Ali Johnson discusses in his extensive review of choreography copyright "Copyrighting TikTok Dances: Choreography in the Internet Age", the challenges in copyright enforcement are not due to a lack of laws but stem from issues related to the fixation requirement and the difficulty in recognizing and attributing dance works. This underscores the necessity for advanced systems capable of accurately identifying and monitoring choreographic content in videos.
Our Dance Recognition system addresses this gap by enabling the global indexing, fixation, tracking, and monitoring of dance IP. Through skeleton-based spatial-temporal embeddings and advanced action recognition methodologies [1], we empower choreographers with powerful copyright protection capabilities. By combining fine-grained analytics with scalable embedding-based recognition technology, we envision a future where every dance movement is identifiable and protected.
At the heart of our framework is a vast and diverse proprietary dataset of dance videos and skeleton movements. This foundation ensures scalability, enabling precise recognition and monitoring of choreographic works across platforms, including social media. Importantly, our system differentiates between commercial and non-commercial uses, fostering a fair ecosystem where creators are protected without stifling the creative freedom of dance challenges and covers.
Research Vision
Choreography ID System: Create an industry-first Choreography ID System, modelled after YouTube’s Content ID, to ensure fair accreditation, monetization, and comprehensive IP management for choreographers in commercial contexts.
Self-Supervised Encoders: Develop robust, self-supervised encoders for choreography recognition, building on Dance2Vec principles to handle fine-grained, heavy motion dance movements [1].
Multi-Dancer Encoding: Extend research into encoding group choreographies, addressing the challenges of spatial dynamics and interaction in multi-dancer performances.
Planned Integrations
MVNT Copyright ID System Beta (Planned for Q4 2025): A comprehensive dance IP management platform, including dance trend analytics, royalty distribution tools, and mechanisms to facilitate the monetization of choreographic works.
MVNT COMMUNITY Onboarding (Planned for 2026): Automatic onboarding of users' dance IP into our systems via the Copyright ID system—synchronising IP metadata with partner organisations like the Korea Choreography Copyright Association (KCCA) to facilitate the institutionalization of dance IP.
Related Work
Wezenberg, A. (2023): "Dance2Vec: Choreography Recognition via Spatial-Temporal Skeleton Embeddings." Proposed a scalable approach for online choreography recognition using skeleton-based spatial-temporal embeddings to elevate copyright fixation from video to embeddings.
Own your movement.
MVNT
Last updated
