open3d.core.HashMap¶

class open3d.core.HashMap¶

A HashMap is an unordered map from key to value wrapped by Tensors.

__init__(*args, **kwargs)¶

Overloaded function.

__init__(self, init_capacity, key_dtype, key_element_shape, value_dtype, value_element_shape, device=CPU:0)

Parameters

init_capacity (int) – Initial capacity of a hash container.
key_dtype (open3d.core.Dtype) – Data type for the input key tensor.
key_element_shape (open3d.core.SizeVector) – Element shape for the input key tensor. E.g. (3) for 3D coordinate keys.
value_dtype (open3d.core.Dtype) – Data type for the input value tensor.
value_element_shape (open3d.core.SizeVector) – Element shape for the input value tensor. E.g. (1) for mapped index.
(open3d.core.Device (device) – 0): Compute device to store and operate on the hash container.
optional – 0): Compute device to store and operate on the hash container.
default=CPU – 0): Compute device to store and operate on the hash container.

__init__(self, init_capacity, key_dtype, key_element_shape, value_dtypes, value_element_shapes, device=CPU:0)

Parameters

init_capacity (int) – Initial capacity of a hash container.
key_dtype (open3d.core.Dtype) – Data type for the input key tensor.
key_element_shape (open3d.core.SizeVector) – Element shape for the input key tensor. E.g. (3) for 3D coordinate keys.
value_dtypes (List[open3d.core.Dtype]) – List of data type for the input value tensors.
value_element_shapes (List[open3d.core.SizeVector]) – List of element shapes for the input value tensors. E.g. ((8,8,8,1), (8,8,8,3)) for mapped weights and RGB colors stored in 8^3 element arrays.
(open3d.core.Device (device) – 0): Compute device to store and operate on the hash container.
optional – 0): Compute device to store and operate on the hash container.
default=CPU – 0): Compute device to store and operate on the hash container.

activate(self, keys)¶

Activate an array of keys stored in Tensors without copying values.

Parameters: keys (open3d.core.Tensor) – Input keys stored in a tensor of shape (N, key_element_shape).
Returns: tuple

active_buf_indices(self: open3d.cpu.pybind.core.HashMap) → open3d.cpu.pybind.core.Tensor¶: Get the buffer indices corresponding to active entries in the hash map.

capacity(self: open3d.cpu.pybind.core.HashMap) → int¶: Get the capacity of the hash map.

clone(self: open3d.cpu.pybind.core.HashMap) → open3d.cpu.pybind.core.HashMap¶: Clone the hash map, including the data structure and the data buffers.

cpu(self: open3d.cpu.pybind.core.HashMap) → open3d.cpu.pybind.core.HashMap¶: Transfer the hash map to CPU. If the hash map is already on CPU, no copy will be performed.

cuda(self, device_id=0)¶

Transfer the hash map to a CUDA device. If the hash map is already on the specified CUDA device, no copy will be performed.

Parameters: device_id (int, optional, default=0) – Target CUDA device ID.
Returns: open3d.core.HashMap

erase(self, keys)¶

Erase an array of keys stored in Tensors.

Parameters: keys (open3d.core.Tensor) – Input keys stored in a tensor of shape (N, key_element_shape).
Returns: open3d.core.Tensor

find(self, keys)¶

Find an array of keys stored in Tensors.

Parameters: keys (open3d.core.Tensor) – Input keys stored in a tensor of shape (N, key_element_shape).
Returns: tuple

insert(*args, **kwargs)¶

Overloaded function.

insert(self, keys, values)
Insert an array of keys and an array of values stored in Tensors.

Parameters

keys (open3d.core.Tensor) – Input keys stored in a tensor of shape (N, key_element_shape).
values (open3d.core.Tensor) – Input values stored in a tensor of shape (N, value_element_shape).

Returns

tuple

insert(self, keys, list_values)
Insert an array of keys and a list of value arrays stored in Tensors.

Parameters

keys (open3d.core.Tensor) – Input keys stored in a tensor of shape (N, key_element_shape).
list_values (List[open3d.core.Tensor]) – List of input values stored in tensors of corresponding shapes.

Returns

tuple

key_tensor(self: open3d.cpu.pybind.core.HashMap) → open3d.cpu.pybind.core.Tensor¶: Get the key tensor stored in the buffer.

static load(file_name)¶

Load a hash map from a .npz file.

Parameters: file_name (str) – File name of the corresponding .npz file.
Returns: open3d.core.HashMap

reserve(self, capacity)¶

Reserve the hash map given the capacity.

Parameters: capacity (int) – New capacity for rehashing.
Returns: None

save(self, file_name)¶

Save the hash map into a .npz file.

Parameters: file_name (str) – File name of the corresponding .npz file.
Returns: None

size(self: open3d.cpu.pybind.core.HashMap) → int¶: Get the size of the hash map.

to(self, device, copy=False)¶

Convert the hash map to a selected device.

Parameters

device (open3d.core.Device) – Compute device to store and operate on the hash container.
copy (bool, optional, default=False) – If true, a new tensor is always created; if false, the copy is avoided when the original tensor already has the targeted dtype.

Returns

open3d.core.HashMap

value_tensor(*args, **kwargs)¶

Overloaded function.

value_tensor(self)
Get the value tensor stored at index 0.

Returns: open3d.core.Tensor

value_tensor(self, value_buffer_id)
Get the value tensor stored at index i

Parameters: value_buffer_id (int) –
Returns: open3d.core.Tensor

value_tensors(self: open3d.cpu.pybind.core.HashMap) → List[open3d.cpu.pybind.core.Tensor]¶: Get the list of value tensors stored in the buffer.