NYUD-v-2

The NYUD-v-2 dataset is a benchmark that contains RGB and depth images along with pixel-wise annotations for various tasks, used to evaluate multi-task dense prediction models.

Papers using NYUD-v-2 (3)

Tsp-transformer: Task-specific Prompts Boosted Transformer For Holistic Scene Understanding2023 · 6 citesFutureDepth: Learning to Predict the Future Improves Video Depth Estimation2024 · 5 citesTask-Instructed Causal Routing of Vision Foundation Models for Multi-Task Learning2026