configs/detection3d/bevformer/bevformer_base_r101_dcn_nuscenes_blancehybrid.py [76:152]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
                pc_range=point_cloud_range,
                num_points_in_pillar=4,
                return_intermediate=False,
                transformerlayers=dict(
                    type='BEVFormerLayer',
                    attn_cfgs=[
                        dict(
                            type='TemporalSelfAttention',
                            embed_dims=embed_dim,
                            num_levels=1),
                        dict(
                            type='SpatialCrossAttention',
                            pc_range=point_cloud_range,
                            deformable_attention=dict(
                                type='MSDeformableAttention3D',
                                embed_dims=embed_dim,
                                num_points=8,
                                num_levels=num_levels),
                            embed_dims=embed_dim,
                        )
                    ],
                    ffn_cfgs=dict(
                        type='FFN',
                        embed_dims=256,
                        feedforward_channels=ffn_dim,
                        num_fcs=2,
                        ffn_drop=0.1,
                        act_cfg=dict(type='ReLU', inplace=True),
                    ),
                    operation_order=('self_attn', 'norm', 'cross_attn', 'norm',
                                     'ffn', 'norm'))),
            decoder=dict(
                type='Detr3DTransformerDecoder',
                num_layers=6,
                return_intermediate=True,
                transformerlayers=dict(
                    type='DetrTransformerDecoderLayer',
                    attn_cfgs=[
                        dict(
                            type='MultiheadAttention',
                            embed_dims=embed_dim,
                            num_heads=8,
                            dropout=0.1),
                        dict(
                            type='CustomMSDeformableAttention',
                            embed_dims=embed_dim,
                            num_levels=1),
                    ],
                    ffn_cfgs=dict(
                        type='FFN',
                        embed_dims=256,
                        feedforward_channels=ffn_dim,
                        num_fcs=2,
                        ffn_drop=0.1,
                        act_cfg=dict(type='ReLU', inplace=True),
                    ),
                    operation_order=('self_attn', 'norm', 'cross_attn', 'norm',
                                     'ffn', 'norm')))),
        bbox_coder=dict(
            type='NMSFreeBBoxCoder',
            post_center_range=[-61.2, -61.2, -10.0, 61.2, 61.2, 10.0],
            pc_range=point_cloud_range,
            max_num=300,
            voxel_size=voxel_size,
            num_classes=10),
        positional_encoding=dict(
            type='LearnedPositionalEncoding',
            num_feats=pos_dim,
            row_num_embed=bev_h,
            col_num_embed=bev_w,
        ),
        loss_cls=dict(
            type='FocalLoss',
            use_sigmoid=True,
            gamma=2.0,
            alpha=0.25,
            loss_weight=2.0),
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



configs/detection3d/bevformer/bevformer_tiny_r50_nuscenes.py [84:160]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
                pc_range=point_cloud_range,
                num_points_in_pillar=4,
                return_intermediate=False,
                transformerlayers=dict(
                    type='BEVFormerLayer',
                    attn_cfgs=[
                        dict(
                            type='TemporalSelfAttention',
                            embed_dims=embed_dim,
                            num_levels=1),
                        dict(
                            type='SpatialCrossAttention',
                            pc_range=point_cloud_range,
                            deformable_attention=dict(
                                type='MSDeformableAttention3D',
                                embed_dims=embed_dim,
                                num_points=8,
                                num_levels=num_levels),
                            embed_dims=embed_dim,
                        )
                    ],
                    ffn_cfgs=dict(
                        type='FFN',
                        embed_dims=256,
                        feedforward_channels=ffn_dim,
                        num_fcs=2,
                        ffn_drop=0.1,
                        act_cfg=dict(type='ReLU', inplace=True),
                    ),
                    operation_order=('self_attn', 'norm', 'cross_attn', 'norm',
                                     'ffn', 'norm'))),
            decoder=dict(
                type='Detr3DTransformerDecoder',
                num_layers=6,
                return_intermediate=True,
                transformerlayers=dict(
                    type='DetrTransformerDecoderLayer',
                    attn_cfgs=[
                        dict(
                            type='MultiheadAttention',
                            embed_dims=embed_dim,
                            num_heads=8,
                            dropout=0.1),
                        dict(
                            type='CustomMSDeformableAttention',
                            embed_dims=embed_dim,
                            num_levels=1),
                    ],
                    ffn_cfgs=dict(
                        type='FFN',
                        embed_dims=256,
                        feedforward_channels=ffn_dim,
                        num_fcs=2,
                        ffn_drop=0.1,
                        act_cfg=dict(type='ReLU', inplace=True),
                    ),
                    operation_order=('self_attn', 'norm', 'cross_attn', 'norm',
                                     'ffn', 'norm')))),
        bbox_coder=dict(
            type='NMSFreeBBoxCoder',
            post_center_range=[-61.2, -61.2, -10.0, 61.2, 61.2, 10.0],
            pc_range=point_cloud_range,
            max_num=300,
            voxel_size=voxel_size,
            num_classes=10),
        positional_encoding=dict(
            type='LearnedPositionalEncoding',
            num_feats=pos_dim,
            row_num_embed=bev_h,
            col_num_embed=bev_w,
        ),
        loss_cls=dict(
            type='FocalLoss',
            use_sigmoid=True,
            gamma=2.0,
            alpha=0.25,
            loss_weight=2.0),
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -