threedb.utils

threedb.utils.str_to_dtype(dtype_str: str) → torch.dtype

class threedb.utils.CyclicBuffer(buffers: Optional[Dict[str, Tuple[List[int], str]]] = None, size: int = 1001, with_tqdm: bool = True)

Bases: object

A concurrent cyclic buffer with reference counting to store the result and avoid copying them to every sub process.

__init__(buffers: Optional[Dict[str, Tuple[List[int], str]]] = None, size: int = 1001, with_tqdm: bool = True) → None

Initialize self. See help(type(self)) for accurate signature.

declare_buffers(buffers: Dict[str, Tuple[List[int], str]]) → bool

free(ind: int, reg_id: int)

register()

process_events()

next_find_index() → int

allocate(data: Dict[str, torch.Tensor])

close()

threedb.utils.overwrite_control(control: threedb.controls.base_control.BaseControl, data: Dict[str, Union[Tuple[float, float], List[Any]]])

threedb.utils.init_control(cfg: Dict[str, Any], root_folder: str)

threedb.utils.init_policy(description)

threedb.utils.load_inference_model(args)