Detect all objects 21.3% better than SAM by expanding prompts for Grounding DINO with DiPEx
Detect all objects 21.3% better than SAM by expanding prompts for Grounding DINO with DiPEx
DiPEx: Dispersing Prompt Expansion for Class-Agnostic Object Detection
arXiv paper abstract https://arxiv.org/abs/2406.14924
arXiv PDF paper https://arxiv.org/pdf/2406.14924
Class-agnostic object detection (OD) … investigate using vision-language models (VLMs) to enhance object detection via a self-supervised prompt learning strategy.
… propose a Dispersing Prompt Expansion (DiPEx) … expand a set of distinct, non-overlapping hyperspherical prompts to enhance recall rates
… DiPEx initiates the process by self-training generic parent prompts and selecting the one with the highest semantic uncertainty for further expansion … child prompts … inherit semantics from their parent prompts while capturing more fine-grained semantics.
… apply dispersion losses to ensure high inter-class discrepancy among child prompts while preserving semantic consistency between parent-child prompt pairs.
To prevent excessive growth of the prompt sets, … utilize the maximum angular coverage (MAC) of the semantic space as a criterion for early termination.
… DiPEx … achieving a 21.3% AP improvement over SAM …
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b
