Test-Time Prototype Adaptation for Open-Vocabulary Semantic Segmentation
This work proposes Test-time Prototype Adaptation (TPA), a training-free plug-in that operates at the output level, leaving the host's forward pass and weights unmodified.
2 papers indexed here
We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.
Not the right person? Other researchers publish under this name.
This work proposes Test-time Prototype Adaptation (TPA), a training-free plug-in that operates at the output level, leaving the host's forward pass and weights unmodified.
Evaluating four representative open-source VLMs on 60 curated rallies via a zero-shot protocol, it is found that models remain weak across all diagnostic levels, with especially clear bottlenecks in spatial state, interaction binding, and terminal evidence.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.