bev-project/configs/nuscenes/seg/camera-bev256d2.yaml

87 lines
1.9 KiB
YAML

model:
encoders:
lidar: null
camera:
backbone:
type: SwinTransformer
embed_dims: 96
depths: [2, 2, 6, 2]
num_heads: [3, 6, 12, 24]
window_size: 7
mlp_ratio: 4
qkv_bias: true
qk_scale: null
drop_rate: 0.
attn_drop_rate: 0.
drop_path_rate: 0.3
patch_norm: true
out_indices: [1, 2, 3]
with_cp: false
convert_weights: true
init_cfg:
type: Pretrained
checkpoint: https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_tiny_patch4_window7_224.pth
neck:
type: GeneralizedLSSFPN
in_channels: [192, 384, 768]
out_channels: 256
start_level: 0
num_outs: 3
norm_cfg:
type: BN2d
requires_grad: true
act_cfg:
type: ReLU
inplace: true
upsample_cfg:
mode: bilinear
align_corners: false
vtransform:
type: LSSTransform
in_channels: 256
out_channels: 80
image_size: ${image_size}
feature_size: ${[image_size[0] // 8, image_size[1] // 8]}
xbound: [-51.2, 51.2, 0.4]
ybound: [-51.2, 51.2, 0.4]
zbound: [-10.0, 10.0, 20.0]
dbound: [1.0, 60.0, 0.5]
downsample: 2
fuser: null
decoder:
backbone:
type: GeneralizedResNet
in_channels: 80
blocks:
- [2, 160, 2]
- [2, 320, 2]
- [2, 640, 1]
neck:
type: LSSFPN
in_indices: [-1, 0]
in_channels: [640, 160]
out_channels: 256
scale_factor: 2
optimizer:
type: AdamW
lr: 1.0e-4
weight_decay: 0.01
paramwise_cfg:
custom_keys:
absolute_pos_embed:
decay_mult: 0
relative_position_bias_table:
decay_mult: 0
optimizer_config:
grad_clip:
max_norm: 35
norm_type: 2
lr_config:
policy: cyclic
momentum_config:
policy: cyclic