Dataset · sayakpaul
RGB-D indoor video sequences from Kinect
sayakpaul/nyu_depth_v2
Classic dataset of paired RGB indoor images and depth maps for depth estimation tasks.
What's inside
The dataset consists of video sequences of various indoor spaces recorded simultaneously using an RGB camera and a Microsoft Kinect depth sensor. The data volume ranges from 1,000 to 10,000 examples with English descriptions.
What you can do with it
- Training and testing monocular depth estimation models
- Validating 3D indoor reconstruction algorithms
- Tasks
- Depth estimation
- Languages
- English
- Size
- 1K–10K rows
- Downloads this month
- 9.3K
- Likes
- 41
Models trained on it
- gabogabogoo/argusDepth estimation 93
- phanerozoic/argusDepth estimation 69
- gabogabogoo/argus-lite-onnxDepth estimation 55
- phanerozoic/argus-liteDepth estimation 50
- Sohaib36/MonoSceneDepth estimation 0
- ibaiGorordo/lap-depth-nyuDepth estimation 0
- phanerozoic/depth-headsDepth estimation 0
- phanerozoic/argus-edgeImage classification 0
How to load it
from datasets import load_dataset
ds = load_dataset("sayakpaul/nyu_depth_v2")