Autobrik commited on
Commit
b37c3f2
·
verified ·
1 Parent(s): 04ec6d3

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +246 -0
README.md ADDED
@@ -0,0 +1,246 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ task_categories:
4
+ - robotics
5
+ tags:
6
+ - LeRobot
7
+ - ecse552
8
+ - ocbench
9
+ - ur5e
10
+ - mujoco
11
+ - simulation
12
+ configs:
13
+ - config_name: default
14
+ data_files: data/*/*.parquet
15
+ ---
16
+
17
+ This dataset was created using [LeRobot](https://github.com/huggingface/lerobot).
18
+
19
+
20
+ <a class="flex" href="https://huggingface.co/spaces/lerobot/visualize_dataset?path=Autobrik/ecse552-ocbench-source-test">
21
+ <img class="block dark:hidden" src="https://huggingface.co/datasets/huggingface/badges/resolve/main/visualize-this-dataset-xl.svg"/>
22
+ <img class="hidden dark:block" src="https://huggingface.co/datasets/huggingface/badges/resolve/main/visualize-this-dataset-xl-dark.svg"/>
23
+ </a>
24
+
25
+
26
+ ## Dataset Description
27
+
28
+ LeRobot copy of `source-deterministic-test-n50-seed50000` from [Autobrik/ecse552-ocbench-block](https://huggingface.co/datasets/Autobrik/ecse552-ocbench-block)
29
+ (exact .npz data and full documentation), for the McGill ECSE 552 project *Learning to
30
+ Manipulate a New Object from Action-Free Videos*
31
+ ([code](https://github.com/BrikHMP18/ecse-552-project), commit `41f93cc`).
32
+
33
+ Scripted-expert episodes in OCBench (UR5e + Robotiq, MuJoCo): object `source` (red 6x6x6 cm cube),
34
+ 50 episodes, 10 Hz, three 224 x 224 cameras (front, side, wrist), 7 normalized actions (six
35
+ joint increments, one gripper increment). Scene `fixed_yaw`, expert `deterministic`.
36
+
37
+ **Held-out source episodes** (own seed range): only for the inverse dynamics model's test error.
38
+
39
+ `observation.state` (joints) and `observation.environment_state` (object pose) are privileged
40
+ simulator state: the project's models use images only. Episode `i` here is
41
+ `episode_{i:05d}.npz` in the source collection (seed in its `manifest.json`). Videos are
42
+ lossy; train from the .npz copy for exact pixels.
43
+
44
+ - **Homepage:** [More Information Needed]
45
+ - **Paper:** [More Information Needed]
46
+ - **License:** mit
47
+
48
+ ## Dataset Structure
49
+
50
+ [meta/info.json](meta/info.json):
51
+ ```json
52
+ {
53
+ "codebase_version": "v3.0",
54
+ "fps": 10,
55
+ "features": {
56
+ "observation.images.front": {
57
+ "dtype": "video",
58
+ "shape": [
59
+ 224,
60
+ 224,
61
+ 3
62
+ ],
63
+ "names": [
64
+ "height",
65
+ "width",
66
+ "channels"
67
+ ],
68
+ "info": {
69
+ "video.height": 224,
70
+ "video.width": 224,
71
+ "video.codec": "av1",
72
+ "video.pix_fmt": "yuv420p",
73
+ "video.fps": 10,
74
+ "video.channels": 3,
75
+ "has_audio": false,
76
+ "video.g": 2,
77
+ "video.crf": 30,
78
+ "video.preset": 12,
79
+ "video.fast_decode": 0,
80
+ "video.video_backend": "pyav",
81
+ "video.extra_options": {},
82
+ "is_depth_map": false
83
+ }
84
+ },
85
+ "observation.images.side": {
86
+ "dtype": "video",
87
+ "shape": [
88
+ 224,
89
+ 224,
90
+ 3
91
+ ],
92
+ "names": [
93
+ "height",
94
+ "width",
95
+ "channels"
96
+ ],
97
+ "info": {
98
+ "video.height": 224,
99
+ "video.width": 224,
100
+ "video.codec": "av1",
101
+ "video.pix_fmt": "yuv420p",
102
+ "video.fps": 10,
103
+ "video.channels": 3,
104
+ "has_audio": false,
105
+ "video.g": 2,
106
+ "video.crf": 30,
107
+ "video.preset": 12,
108
+ "video.fast_decode": 0,
109
+ "video.video_backend": "pyav",
110
+ "video.extra_options": {},
111
+ "is_depth_map": false
112
+ }
113
+ },
114
+ "observation.images.wrist": {
115
+ "dtype": "video",
116
+ "shape": [
117
+ 224,
118
+ 224,
119
+ 3
120
+ ],
121
+ "names": [
122
+ "height",
123
+ "width",
124
+ "channels"
125
+ ],
126
+ "info": {
127
+ "video.height": 224,
128
+ "video.width": 224,
129
+ "video.codec": "av1",
130
+ "video.pix_fmt": "yuv420p",
131
+ "video.fps": 10,
132
+ "video.channels": 3,
133
+ "has_audio": false,
134
+ "video.g": 2,
135
+ "video.crf": 30,
136
+ "video.preset": 12,
137
+ "video.fast_decode": 0,
138
+ "video.video_backend": "pyav",
139
+ "video.extra_options": {},
140
+ "is_depth_map": false
141
+ }
142
+ },
143
+ "observation.state": {
144
+ "dtype": "float32",
145
+ "shape": [
146
+ 7
147
+ ],
148
+ "names": [
149
+ "shoulder_pan",
150
+ "shoulder_lift",
151
+ "elbow",
152
+ "wrist_1",
153
+ "wrist_2",
154
+ "wrist_3",
155
+ "gripper_driver"
156
+ ]
157
+ },
158
+ "observation.environment_state": {
159
+ "dtype": "float32",
160
+ "shape": [
161
+ 7
162
+ ],
163
+ "names": [
164
+ "object_x",
165
+ "object_y",
166
+ "object_z",
167
+ "object_qw",
168
+ "object_qx",
169
+ "object_qy",
170
+ "object_qz"
171
+ ]
172
+ },
173
+ "action": {
174
+ "dtype": "float32",
175
+ "shape": [
176
+ 7
177
+ ],
178
+ "names": [
179
+ "d_shoulder_pan",
180
+ "d_shoulder_lift",
181
+ "d_elbow",
182
+ "d_wrist_1",
183
+ "d_wrist_2",
184
+ "d_wrist_3",
185
+ "d_gripper"
186
+ ]
187
+ },
188
+ "timestamp": {
189
+ "dtype": "float32",
190
+ "shape": [
191
+ 1
192
+ ],
193
+ "names": null
194
+ },
195
+ "frame_index": {
196
+ "dtype": "int64",
197
+ "shape": [
198
+ 1
199
+ ],
200
+ "names": null
201
+ },
202
+ "episode_index": {
203
+ "dtype": "int64",
204
+ "shape": [
205
+ 1
206
+ ],
207
+ "names": null
208
+ },
209
+ "index": {
210
+ "dtype": "int64",
211
+ "shape": [
212
+ 1
213
+ ],
214
+ "names": null
215
+ },
216
+ "task_index": {
217
+ "dtype": "int64",
218
+ "shape": [
219
+ 1
220
+ ],
221
+ "names": null
222
+ }
223
+ },
224
+ "total_episodes": 50,
225
+ "total_frames": 3818,
226
+ "total_tasks": 1,
227
+ "chunks_size": 1000,
228
+ "data_files_size_in_mb": 100,
229
+ "video_files_size_in_mb": 200,
230
+ "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
231
+ "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4",
232
+ "robot_type": "ur5e_robotiq",
233
+ "splits": {
234
+ "train": "0:50"
235
+ }
236
+ }
237
+ ```
238
+
239
+
240
+ ## Citation
241
+
242
+ **BibTeX:**
243
+
244
+ ```bibtex
245
+ [More Information Needed]
246
+ ```