--- dataset_info: features: - name: image dtype: image - name: depth dtype: image - name: label dtype: image splits: - name: train num_bytes: 152339736 num_examples: 795 - name: test num_bytes: 124128749 num_examples: 654 download_size: 276518209 dataset_size: 276468485 configs: - config_name: default data_files: - split: train path: data/train-* - split: test path: data/test-* task_categories: - image-segmentation --- # NYUDepth v2 Easier version for semantic segmentation. Comes from: Indoor Segmentation and Support Inference from RGBD Images (ECCV 2012) ([PDF](https://cs.nyu.edu/~fergus/datasets/indoor_seg_support.pdf)) ([Bib](https://cs.nyu.edu/~fergus/datasets/indoor_seg_support.bib))