Hugging Face's logo

Datasets:
cyberorigin
/
cyber_twist_the_tube

Modalities:
Tabular
Text
Formats:
json
Size:
< 1K
Libraries:
Datasets
Dask
License:
Dataset card Viewer Files Community
Dataset Viewer
Auto-converted to Parquet
Full Screen
token_dtype
stringclasses
1 value
s
int64
16
16
h
int64
16
16
w
int64
16
16
vocab_size
int64
262k
262k
hz
int64
30
30
tokenizer_ckpt
stringclasses
2 values
num_images
int64
94.9k
425k
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
371,261
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
344,789
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
357,772
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
329,398
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
259,526
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
292,590
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
258,855
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
277,188
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
140,268
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
144,385
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
136,429
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
363,298
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
316,457
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
200,611
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
401,056
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
190,631
uint32
16
16
16
262,144
30
data/magvit2.ckpt
94,891
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
276,060
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
425,320
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
398,222
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
315,581
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
253,638
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
251,936
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
245,124
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
259,909
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
281,718
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
142,346

CyberOrigin Dataset

Our data includes information from home services, the logistics industry, and laboratory scenarios. For more details, please refer to our Offical Data Website

contents of the dataset:

cyber_twist_the_tube # dataset root path
  └── data/
      ├── metadata_ID1_240808.json
      ├── segment_ids_ID1_240808.bin # for each frame segment_ids uniquely points to the segment index that frame i came from. You may want to use this to separate non-contiguous frames from different videos (transitions).
      ├── videos_ID1_240808.bin # 16x16 image patches at 30hz, each patch is vector-quantized into 2^18 possible integer values. These can be decoded into 256x256 RGB images using the provided magvit2.ckpt weights.
      ├── ...
  └── ...

{
    "task": "Twist the Tube",
    "total_episodes": 26258,
    "total_frames": 7329259,
    "token_dtype": "uint32",
    "vocab_size": 262144,
    "fps": 30,
    "manipulation_type": "Bi-Manual",
    "language_annotation": "None",
    "scene_type": "Table Top",
    "data_collect_method": "Directly Collection on Human"
}
Downloads last month
45
Size of the auto-converted Parquet files:
65.3 kB
Number of rows:
27