@misc{indiciae3f4a7b11ccf5,
  title = {SCI-CLIP: Segment-Centric Inference with Reference Memory for Training-Free Open-Vocabulary Segmentation},
  author = {Mohamad Zamini and Diksha Shukla},
  year = {2026},
  url = {https://arxiv.org/abs/2608.05627},
  note = {Source identifier: 2608.05627}
}