Controllable Image Captioning with Prompt-Conditioned Scene Rewards
FoCUS is presented, a controllable image captioning method that lets users steer captions toward specific semantic emphases through natural-language control prompts, and SCoPE, a benchmark with contrastive Include/Avoid constraints for measuring both target content coverage and out-of-scope suppression.
Jongyeop Hyun, Taeyoung Kim, Hyounghun Kim
· 0 citations