Human action understanding method based on large language model and adaptive contrast decoding

By using the FLARE framework and the AdaCD adaptive contrastive decoding method, combined with a large language model and an action encoder, the problems of poor generalization ability and illusion in existing technologies are solved, thereby improving the accuracy and efficiency of human action understanding.

CN119514683BActive Publication Date: 2025-12-26XIDIAN UNIV
2 Cites 0 Cited by

Patent Information

Application Number
CN202411534652.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-10-31
Publication Date
2025-12-26
Estimated Expiration
2044-10-31

AI Technical Summary

Technical Problem

Existing methods for understanding human actions lack generalization ability across different tasks, and their over-reliance on the language priors of frozen large language models leads to illusion problems, resulting in inaccurate understanding of human actions.

Method used

We employ a human action understanding method based on a large language model and adaptive contrastive decoding. By combining a frozen autoregressive large-scale language model (LLM) and a pre-trained action encoder through the FLARE framework, and combining it with the adaptive contrastive decoding method AdaCD, we adjust the weights of relevant tags, optimize model performance, reduce over-reliance on language priors, and improve the accuracy and generalization ability of action understanding.

Benefits of technology

It improves the generalization ability of human action understanding methods, reduces computational resource consumption, reduces the illusion problem, and achieves accurate understanding of various action understanding tasks.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The application belongs to the technical field of artificial intelligence, and discloses a human action understanding method based on a large language model and adaptive contrast decoding, proposes a FLARE framework, introduces a pre-trained frozen large language model LLM and a frozen action encoder, aligns the action and language semantics, and uses an adaptive contrast decoding AdaCD method to reduce excessive dependence on language prior, thereby improving the model's understanding ability of the action, enabling the method to handle various action understanding tasks, improving the generalization ability of the human action understanding method, and reducing the task specificity and computing resource consumption problems in the prior art.
Need to check novelty before this filing date? Find Prior Art