chophilip21/VLM-cross-scene-portrait-matcher

Title: VLM-cross-scene-portrait-matcher

Description: cross-scene subject alignment framework. Use pretrained vision-language model to propose candidate individuals in crowded images, extracting face embeddings for both scenes, and iteratively matching and filtering these embeddings to align subjects from a noisy, unordered set of group photos to a reference set of single‐subject portraits

GitHub Organization: https://github.com/chophilip21/VLM-cross-scene-portrait-matcher

Schema URL: https://chophilip21.github.io/VLM-cross-scene-portrait-matcher/

Domain: Computer Science

Status: unknown

Contacts: chophilip21