(self, conv_out, segSize=None)
| 197 | ) |
| 198 | |
| 199 | def forward(self, conv_out, segSize=None): |
| 200 | conv4 = conv_out[-2] |
| 201 | conv5 = conv_out[-1] |
| 202 | ###### |
| 203 | x1 = nn.functional.interpolate(conv4, size=segSize, mode='bilinear', align_corners=False) |
| 204 | ###### |
| 205 | input_size = conv5.size() |
| 206 | ppm_out = [conv5] |
| 207 | for pool_scale in self.ppm: |
| 208 | ppm_out.append(nn.functional.interpolate( |
| 209 | pool_scale(conv5), |
| 210 | (input_size[2], input_size[3]), |
| 211 | mode='bilinear', align_corners=False)) |
| 212 | ppm_out = torch.cat(ppm_out, 1) |
| 213 | x = self.conv_last(ppm_out) |
| 214 | x = nn.functional.interpolate( |
| 215 | x, size=segSize, mode='bilinear', align_corners=False) |
| 216 | x = nn.functional.softmax(x, dim=1) |
| 217 | return x, x1 |
| 218 | |
| 219 |
nothing calls this directly
no outgoing calls
no test coverage detected