Datasets
Standard Dataset
3D-SPAD
- Citation Author(s):
- Submitted by:
- Fei Hu
- Last updated:
- Fri, 01/12/2024 - 21:34
- DOI:
- 10.21227/y0at-va95
- License:
73 Views- Categories:
- Keywords:
Abstract
The 3D-SPAD dataset we proposed is preprocessed and stored in the JSON file. If you want to visualize, you can visit this webpage to get the original 3D scene data.
Our dataset contains three scenes: 'bedroom','bathroom','living room'. The object types include:'fish tank','bathub','laptop','office chair','desk','changer','tower rack','single bed','table lamp','double bed','armchair','computer','dresser','glass','plant','floor lamp','tv stand','vase','books','television','corrector','sink','washer','ottoman','sof','piano','bunker bed'.'dressing table','stand','mirror','shelving','fruit bowl', 'coffee table', 'clock', 'toilet', 'chair', 'xbox', 'wardrobe cabinet', 'wall lamp', 'shower', 'bottle', 'cup', 'stereo set', 'wall', 'rug', 'chessboard'
Data structure(Json)：
-scene_name: Scene ID.
-zero_num: Number of nodes with empty label. (from the end)
-feature: Node eigenvector matrix.
-adj: adjacency matrix.
-gmm: Gaussian mixture model output probability matrix.
-supported_mask: Invalid matrix.
-zero_mask: Invalid matrix.
-evaluate: Object label.
-scene_score: Scene label.
