Patent Application for Improved Video Prediction Technology Filed

Strategic Insights -

Researchers Norimasa Kobori, Quan Kong, Jingjing Pan, and Rajat Saini have filed a patent application for an improved video prediction technology that generates text information by predicting the contents of videos. The invention aims to address limitations in conventional technology, which fails to provide accurate verbalization of positions, behaviors, and gazes of people, objects, or motion changes on time series and geometrical information.

Key Takeaways:

  • The patent application, filed on April 7, 2025, describes an information processing apparatus that includes a controller configured to input a video of a first predetermined period to a temporal direction encoder to acquire a first temporal feature amount.
  • The apparatus also extracts instance information from the video and inputs it to a spatial direction encoder to acquire a spatial feature amount.
  • The first temporal feature amount undergoes a cross-attention operation based on language information of a second predetermined period to acquire a second temporal feature amount.
  • The second temporal feature amount and the spatial feature amount are then input to a language processing model to output text information corresponding to the video of the first predetermined period.
  • The claims outline four variations of the information processing apparatus, method, and non-transitory computer-readable medium.
  • Implementation details specify the use of temporal and spatial direction encoders, as well as a language processing model.

Statistics:

  • The patent application has been filed with the U.S. Patent and Trademark Office and was made available online on October 16, 2025.
  • The application includes four claims, detailing the apparatus, method, and computer-readable medium for improved video prediction technology.
  • The invention aims to overcome limitations in conventional technology, which fails to provide accurate verbalization of positions, behaviors, and gazes of people, objects, or motion changes on time series and geometrical information.

Sources:

  • Kobori, N., Kong, Q., Pan, J., & Saini, R. Information Processing Apparatus, Method, And Non-Transitory Computer Readable Medium. U.S. Patent Application Number 20250322658, filed April 7, 2025 and posted October 16, 2025. Patent URL: https://ppubs.uspto.gov/pubwebapp/external.html?q=(20250322658)&db=US-PGPUB&type=ids.