"""An attention module used in BEVFormer based on Deformable-Detr."""
from __future__ import annotations
import math
import torch
from torch import Tensor, nn
from vis4d.op.layer.ms_deform_attn import (
MSDeformAttentionFunction,
is_power_of_2,
ms_deformable_attention_cpu,
)
from vis4d.op.layer.weight_init import constant_init, xavier_init
[docs]
class TemporalSelfAttention(nn.Module):
"""Temperal Self Attention."""
def __init__(
self,
embed_dims: int = 256,
num_heads: int = 8,
num_levels: int = 4,
num_points: int = 4,
num_bev_queue: int = 2,
im2col_step: int = 64,
dropout: float = 0.1,
batch_first: bool = True,
) -> None:
"""Init.
Args:
embed_dims (int): The embedding dimension of Attention. Default:
256.
num_heads (int): Parallel attention heads. Default: 64.
num_levels (int): The number of feature map used in Attention.
Default: 4.
num_points (int): The number of sampling points for each query in
each head. Default: 4.
num_bev_queue (int): In this version, we only use one history BEV
and one currenct BEV. The length of BEV queue is 2.
im2col_step (int): The step used in image_to_column. Default: 64.
dropout (float): A Dropout layer on `inp_identity`. Default: 0.1.
batch_first (bool): Key, Query and Value are shape of (batch, n,
embed_dim) or (n, batch, embed_dim). Default to True.
"""
super().__init__()
if embed_dims % num_heads != 0:
raise ValueError(
f"embed_dims must be divisible by num_heads, "
f"but got {embed_dims} and {num_heads}"
)
is_power_of_2(embed_dims // num_heads)
self.dropout = nn.Dropout(dropout)
self.batch_first = batch_first
self.im2col_step = im2col_step
self.embed_dims = embed_dims
self.num_levels = num_levels
self.num_heads = num_heads
self.num_points = num_points
self.num_bev_queue = num_bev_queue
self.sampling_offsets = nn.Linear(
embed_dims * self.num_bev_queue,
num_bev_queue * num_heads * num_levels * num_points * 2,
)
self.attention_weights = nn.Linear(
embed_dims * self.num_bev_queue,
num_bev_queue * num_heads * num_levels * num_points,
)
self.value_proj = nn.Linear(embed_dims, embed_dims)
self.output_proj = nn.Linear(embed_dims, embed_dims)
self.init_weights()
[docs]
def init_weights(self) -> None:
"""Default initialization for Parameters of Module."""
constant_init(self.sampling_offsets, 0.0)
thetas = torch.mul(
torch.arange(self.num_heads, dtype=torch.float32),
(2.0 * math.pi / self.num_heads),
)
grid_init = torch.stack([thetas.cos(), thetas.sin()], -1)
grid_init = (
(grid_init / grid_init.abs().max(-1, keepdim=True)[0])
.view(self.num_heads, 1, 1, 2)
.repeat(
1, self.num_levels * self.num_bev_queue, self.num_points, 1
)
)
for i in range(self.num_points):
grid_init[:, :, i, :] *= i + 1
self.sampling_offsets.bias.data = grid_init.view(-1)
constant_init(self.attention_weights, val=0.0, bias=0.0)
xavier_init(self.value_proj, distribution="uniform", bias=0.0)
xavier_init(self.output_proj, distribution="uniform", bias=0.0)
[docs]
def forward(
self,
query: Tensor,
reference_points: Tensor,
value: Tensor | None,
spatial_shapes: Tensor,
level_start_index: Tensor,
key_padding_mask: Tensor | None = None,
identity: Tensor | None = None,
query_pos: Tensor | None = None,
) -> Tensor:
"""Forward Function of MultiScaleDeformAttention.
Args:
query (Tensor): Query of Transformer with shape (num_query, bs,
embed_dims).
reference_points (Tensor): The normalized reference points with
shape (bs, num_query, num_levels, 2), all elements is range in
[0, 1], top-left (0,0), bottom-right (1, 1), including padding
area. or (N, Length_{query}, num_levels, 4), add additional two
dimensions is (w, h) to form reference boxes.
value (Tensor): The value tensor with shape (num_key, bs,
embed_dims).
spatial_shapes (Tensor): Spatial shape of features in different
levels. With shape (num_levels, 2), last dimension represents
(h, w).
level_start_index (Tensor): The start index of each level.
A tensor has shape ``(num_levels, )`` and can be represented
as [0, h_0*w_0, h_0*w_0+h_1*w_1, ...].
key_padding_mask (Tensor): ByteTensor for value, with shape [bs,
num_key].
identity (Tensor): The tensor used for addition, with the
same shape as query. Default None. If None, query will be used.
query_pos (Tensor, optional): The positional encoding for query.
Default: None.
Returns:
Tensor: forwarded results with shape [num_query, bs, embed_dims].
"""
if value is None:
assert self.batch_first
bs, len_bev, c = query.shape
value = torch.stack([query, query], 1).reshape(bs * 2, len_bev, c)
if identity is None:
identity = query
if query_pos is not None:
query = query + query_pos
if not self.batch_first:
# change to (bs, num_query ,embed_dims)
query = query.permute(1, 0, 2)
value = value.permute(1, 0, 2)
bs, num_query, embed_dims = query.shape
_, num_value, _ = value.shape
assert (spatial_shapes[:, 0] * spatial_shapes[:, 1]).sum() == num_value
assert self.num_bev_queue == 2
query = torch.cat([value[:bs], query], -1)
value = self.value_proj(value)
assert isinstance(value, Tensor)
if key_padding_mask is not None:
value = value.masked_fill(key_padding_mask[..., None], 0.0)
value = value.reshape(
bs * self.num_bev_queue, num_value, self.num_heads, -1
)
sampling_offsets = self.sampling_offsets(query)
sampling_offsets = sampling_offsets.view(
bs,
num_query,
self.num_heads,
self.num_bev_queue,
self.num_levels,
self.num_points,
2,
)
attention_weights = self.attention_weights(query).view(
bs,
num_query,
self.num_heads,
self.num_bev_queue,
self.num_levels * self.num_points,
)
attention_weights = attention_weights.softmax(-1)
attention_weights = attention_weights.view(
bs,
num_query,
self.num_heads,
self.num_bev_queue,
self.num_levels,
self.num_points,
)
attention_weights = (
attention_weights.permute(0, 3, 1, 2, 4, 5)
.reshape(
bs * self.num_bev_queue,
num_query,
self.num_heads,
self.num_levels,
self.num_points,
)
.contiguous()
)
sampling_offsets = sampling_offsets.permute(
0, 3, 1, 2, 4, 5, 6
).reshape(
bs * self.num_bev_queue,
num_query,
self.num_heads,
self.num_levels,
self.num_points,
2,
)
if reference_points.shape[-1] == 2:
offset_normalizer = torch.stack(
[spatial_shapes[..., 1], spatial_shapes[..., 0]], -1
)
sampling_locations = (
reference_points[:, :, None, :, None, :]
+ sampling_offsets
/ offset_normalizer[None, None, None, :, None, :]
)
elif reference_points.shape[-1] == 4:
sampling_locations = (
reference_points[:, :, None, :, None, :2]
+ sampling_offsets
/ self.num_points
* reference_points[:, :, None, :, None, 2:]
* 0.5
)
else:
raise ValueError(
f"Last dim of reference_points must be"
f" 2 or 4, but get {reference_points.shape[-1]} instead."
)
if torch.cuda.is_available() and value.is_cuda:
output = MSDeformAttentionFunction.apply(
value,
spatial_shapes,
level_start_index,
sampling_locations,
attention_weights,
self.im2col_step,
)
else:
output = ms_deformable_attention_cpu(
value,
spatial_shapes,
sampling_locations,
attention_weights,
)
# output shape (bs*num_bev_queue, num_query, embed_dims)
# (bs*num_bev_queue, num_query, embed_dims)
# -> (num_query, embed_dims, bs*num_bev_queue)
output = output.permute(1, 2, 0)
# fuse history value and current value
# (num_query, embed_dims, bs*num_bev_queue)
# -> (num_query, embed_dims, bs, num_bev_queue)
output = output.view(num_query, embed_dims, bs, self.num_bev_queue)
output = output.mean(-1)
# (num_query, embed_dims, bs)-> (bs, num_query, embed_dims)
output = output.permute(2, 0, 1)
output = self.output_proj(output)
if not self.batch_first:
output = output.permute(1, 0, 2)
return self.dropout(output) + identity