Initialize layers. Args: in_channels (tuple[int]): Number of channels in input tensors. out_channels (int): Number of channels in the neck output tensors. num_classes (int): Number of classes.
(self, in_channels: Tuple[int], out_channels: int,
num_classes: int)
| 100 | ME.MinkowskiBatchNorm(out_channels), ME.MinkowskiELU()) |
| 101 | |
| 102 | def _init_layers(self, in_channels: Tuple[int], out_channels: int, |
| 103 | num_classes: int): |
| 104 | """Initialize layers. |
| 105 | |
| 106 | Args: |
| 107 | in_channels (tuple[int]): Number of channels in input tensors. |
| 108 | out_channels (int): Number of channels in the neck output tensors. |
| 109 | num_classes (int): Number of classes. |
| 110 | """ |
| 111 | # neck layers |
| 112 | self.pruning = ME.MinkowskiPruning() |
| 113 | for i in range(len(in_channels)): |
| 114 | if i > 0: |
| 115 | self.__setattr__( |
| 116 | f'up_block_{i}', |
| 117 | self._make_up_block(in_channels[i], in_channels[i - 1])) |
| 118 | self.__setattr__(f'out_block_{i}', |
| 119 | self._make_block(in_channels[i], out_channels)) |
| 120 | |
| 121 | # head layers |
| 122 | self.conv_cls = ME.MinkowskiConvolution(out_channels, |
| 123 | num_classes, |
| 124 | kernel_size=1, |
| 125 | bias=True, |
| 126 | dimension=3) |
| 127 | |
| 128 | def init_weights(self): |
| 129 | """Initialize weights.""" |
no test coverage detected