Dimensions overall score 8.0

GitHub Code Pulse

No public code linked for this paper yet.

• CAPT is a Confusion-Aware Prompt Tuning framework designed to reduce vision-language misalignment.method 70%
• CAPT constructs a Confusion Bank to explicitly model stable confusion relationships across categories and misclassified samples.method 70%
• CAPT utilizes a Semantic Confusion Miner (SEM) to capture global inter-class confusion through semantic difference and commonality prompts.method 70%
• CAPT employs a Sample Confusion Miner (SAM) with a Diff-Manner Adapter to capture sample-level confusion cues by integrating global and local contexts.method 70%
• A Multi-Granularity Difference Expert (MGDE) module is designed to jointly leverage semantic- and sample-level experts for confusion-aware reasoning.method 70%
• CAPT significantly reduces confusion-induced errors and enhances discriminability and generalization of both base and novel classes.result 70%
• CAPT successfully resolves 50.72 percent of confusable sample pairs.result 70%
• The effectiveness of CAPT relies on previously identified confusion patterns, which may require adjustments for new data or categories.limitation 70%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card