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
1 value
num_images
int64
7.64k
504k
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
451,010
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
142,409
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
504,486
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
502,831
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
462,158
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
406,734
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
7,638
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
171,681
uint32
16
16
16
262,144
30
imagenet_256_L.ckpt
302,218

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_fold_towels # 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": "Fold Towels",
    "total_episodes": 6927,
    "total_frames": 2951165,
    "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
115
Edit dataset card