TY  - RPRT
TI  - MIRCaps: A Large-Scale Mixed-Domain Dataset with Image-Level and Region-Level Captions for Fine-Grained Vision-Language Learning
AU  - Arlindo Luciano Tulumba Roberto
AU  - Hyungjoon Kim
PY  - 2026
UR  - https://arxiv.org/abs/2606.21419
ID  - 2606.21419
ER  -