PRIST, a benchmark for pixel-level segmentation through multi-turn conversations, and the MIRAS model achieve the best reported scores on this new task.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Pixel-Level Reasoning Segmentation via Multi-turn Conversations
PRIST, a benchmark for pixel-level segmentation through multi-turn conversations, and the MIRAS model achieve the best reported scores on this new task.