bpiyush commited on
Commit
4ecc66e
β€’
1 Parent(s): ff08603

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +157 -0
README.md ADDED
@@ -0,0 +1,157 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ # The Sound of Water: Inferring Physical Properties from Pouring Liquids
3
+
4
+ This dataset is associated with the paper "The Sound of Water: Inferring Physical Properties from Pouring Liquids".
5
+
6
+ TODO: add citation and paper details.
7
+
8
+ The dataset is stored in the following directory structure:
9
+ ```sh
10
+ SoundOfWater/
11
+ |-- annotations
12
+ |-- assets
13
+ |-- audios
14
+ |-- README.md
15
+ |-- splits
16
+ |-- videos
17
+ `-- youtube_samples
18
+
19
+ 6 directories, 1 file
20
+ ```
21
+
22
+ ## Table of Contents
23
+
24
+ - [The Sound of Water: Inferring Physical Properties from Pouring Liquids](#the-sound-of-water-inferring-physical-properties-from-pouring-liquids)
25
+ - [Table of Contents](#table-of-contents)
26
+ - [Dataset Overview](#dataset-overview)
27
+ - [Video and audio samples](#video-and-audio-samples)
28
+ - [Splits](#splits)
29
+ - [Annotations](#annotations)
30
+ - [Container measurements and other metadata](#container-measurements-and-other-metadata)
31
+ - [Container bounding boxes](#container-bounding-boxes)
32
+ - [YouTube samples](#youtube-samples)
33
+ - [Citation](#citation)
34
+
35
+
36
+ ## Dataset Overview
37
+
38
+
39
+ ## Video and audio samples
40
+
41
+ The video and audio samples are stored in the `./videos/` and `./audios/` directories, respectively.
42
+ Note that we have trimmed the videos between the precise start and end of the pouring action.
43
+ If you need untrimmed videos, please contact us separately and we may be able to help.
44
+
45
+ The metadata for each video is a row in "./annotations/localisation.csv".
46
+
47
+ ## Splits
48
+
49
+ We create four splits of the dataset.
50
+ All of the splits can be found in the `./splits/` directory.
51
+ The splits are as follows:
52
+ <table>
53
+ <style>
54
+ table td:nth-child(n+2), table th:nth-child(n+2) {
55
+ text-align: center;
56
+ }
57
+ </style>
58
+ <tr>
59
+ <th>Split</th>
60
+ <th colspan="2">Opacity</th>
61
+ <th colspan="3">Shapes</th>
62
+ <th>Containers</th>
63
+ <th>Videos</th>
64
+ <th>Description</th>
65
+ </tr>
66
+ <tr>
67
+ <td></td>
68
+ <td><i>Transparent</i></td>
69
+ <td><i>Opaque</i></td>
70
+ <td><i>Cylinder</i></td>
71
+ <td><i>Semi-cone</i></td>
72
+ <td><i>Bottle</i></td>
73
+ <td></td>
74
+ <td></td>
75
+ <td></td>
76
+ </tr>
77
+ <tr>
78
+ <td>Train</td>
79
+ <td>βœ“</td>
80
+ <td>βœ—</td>
81
+ <td>βœ“</td>
82
+ <td>βœ“</td>
83
+ <td>βœ—</td>
84
+ <td>18</td>
85
+ <td>195</td>
86
+ <td>Transparent cylinder-like containers</td>
87
+ </tr>
88
+ <tr>
89
+ <td>Test I</td>
90
+ <td>βœ“</td>
91
+ <td>βœ—</td>
92
+ <td>βœ“</td>
93
+ <td>βœ“</td>
94
+ <td>βœ—</td>
95
+ <td>13</td>
96
+ <td>54</td>
97
+ <td>Test set with seen containers</td>
98
+ </tr>
99
+ <tr>
100
+ <td>Test II</td>
101
+ <td>βœ—</td>
102
+ <td>βœ“</td>
103
+ <td>βœ“</td>
104
+ <td>βœ“</td>
105
+ <td>βœ—</td>
106
+ <td>19</td>
107
+ <td>327</td>
108
+ <td>Test set with unseen containers</td>
109
+ </tr>
110
+ <tr>
111
+ <td>Test III</td>
112
+ <td>βœ“</td>
113
+ <td>βœ“</td>
114
+ <td>βœ“</td>
115
+ <td>βœ“</td>
116
+ <td>βœ“</td>
117
+ <td>25</td>
118
+ <td>434</td>
119
+ <td>Shape clf. with unseen containers</td>
120
+ </tr>
121
+ </table>
122
+
123
+
124
+ TODO: add test_III.txt file.
125
+
126
+ ## Annotations
127
+
128
+ The set of containers used is shown in this image:
129
+
130
+ <figure>
131
+ <img src="./assets/container_canvas.png" alt="Containers used in the dataset" style="width:70%">
132
+ <figcaption>Containers used in the dataset</figcaption>
133
+ </figure>
134
+
135
+ #### Container measurements and other metadata
136
+
137
+ All metadata for the containers is stored in the `./annotations/` file.
138
+
139
+ | **File** | **Description** |
140
+ | --- | --- |
141
+ | `localisation.csv` | Each row is metadata (e.g., container) for each video. |
142
+ | `containers.yaml` | Metadata for each container. |
143
+ | `liquids.yaml` | Metadata for each liquid. |
144
+ | `materials.yaml` | Metadata for each material. |
145
+
146
+
147
+ #### Container bounding boxes
148
+
149
+ The bounding box annotations for containers are stored here: `./annotations/container_bboxes/`.
150
+ These are generated in a zero-shot manner using [LangSAM](https://github.com/luca-medeiros/lang-segment-anything).
151
+
152
+
153
+ ## YouTube samples
154
+
155
+ We also provide 4 samples searched from YouTube. These are used for qualitative evaluation.
156
+
157
+ ## Citation