Python torch.nn.functional.avg_pool3d() Examples

The following are 30 code examples of torch.nn.functional.avg_pool3d(). You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may also want to check out all available functions/classes of the module torch.nn.functional , or try the search function .
Example #1
Source File: dpn3d.py    From DeepLung with GNU General Public License v3.0 6 votes vote down vote up 
def forward(self, x):
        if debug: print '0', x.size(), 64
        out = F.relu(self.bn1(self.conv1(x)))
        if debug: print '1', out.size()
        out = self.layer1(out)
        if debug: print '2', out.size()
        out = self.layer2(out)
        if debug: print '3', out.size()
        out = self.layer3(out)
        if debug: print '4', out.size()
        out = self.layer4(out)
        if debug: print '5', out.size()
        out = F.avg_pool3d(out, 4)
        if debug: print '6', out.size()
        out_1 = out.view(out.size(0), -1)
        if debug: print '7', out_1.size()
        out = self.linear(out_1)
        if debug: print '8', out.size()
        return out, out_1
Example #2
Source File: model_3d_lc.py    From DPC with MIT License 6 votes vote down vote up 
def forward(self, block):
        # seq1: [B, N, C, SL, W, H]
        (B, N, C, SL, H, W) = block.shape
        block = block.view(B*N, C, SL, H, W)
        feature = self.backbone(block)
        del block 
        feature = F.relu(feature)
        
        feature = F.avg_pool3d(feature, (self.last_duration, 1, 1), stride=1)
        feature = feature.view(B, N, self.param['feature_size'], self.last_size, self.last_size) # [B*N,D,last_size,last_size]
        context, _ = self.agg(feature)
        context = context[:,-1,:].unsqueeze(1)
        context = F.avg_pool3d(context, (1, self.last_size, self.last_size), stride=1).squeeze(-1).squeeze(-1)
        del feature

        context = self.final_bn(context.transpose(-1,-2)).transpose(-1,-2) # [B,N,C] -> [B,C,N] -> BN() -> [B,N,C], because BN operates on id=1 channel.
        output = self.final_fc(context).view(B, -1, self.num_class)

        return output, context
Example #3
Source File: stn.py    From istn with Apache License 2.0 6 votes vote down vote up 
def forward(self, x):
        b, c, d, h, w = x.shape
        xs = F.avg_pool3d(F.relu(self.conv1(x)), 2)
        xs = F.avg_pool3d(F.relu(self.conv2(xs)), 2)
        xs = F.avg_pool3d(F.relu(self.conv3(xs)), 2)
        xs = xs.view(xs.size(0), -1)
        self.regularisation_loss = 30.0 * torch.mean(torch.abs(xs))
        
        # cap the displacement field by max_disp
        xs = torch.tanh(self.fc(xs)) * self.max_disp
        xs = xs.view(-1, *self.cp_grid_shape)

        self.displacement_field = self.compute_displacement(xs) + self.gen_3d_mesh_grid(d, h, w).unsqueeze(0)
        
        # extract first channel for warping
        img = x.narrow(dim=1, start=0, length=1)

        # warp image
        return self.warp_image(img).to(self.device)
Example #4
Source File: models.py    From OBELISK with MIT License 6 votes vote down vote up 
def forward(self, inputImg, sample_grid=None):
    
        B,C,D,H,W = inputImg.size()
        if(sample_grid is None):
            sample_grid = self.sample_grid1
        sample_grid = sample_grid.to(inputImg.device)    
        #pre-smooth image (has to be done in advance for original models )
        #x00 = F.avg_pool3d(inputImg,3,padding=1,stride=1)
        
        _,D_grid,H_grid,W_grid,_ = sample_grid.size()
        input = F.grid_sample(inputImg, (sample_grid.view(1,1,-1,1,3).repeat(B,1,1,1,1) + self.offset1[:,:,:,0:1,:])).view(B,-1,D_grid,H_grid,W_grid)-\
        F.grid_sample(inputImg, (sample_grid.view(1,1,-1,1,3).repeat(B,1,1,1,1) + self.offset1[:,:,:,1:2,:])).view(B,-1,D_grid,H_grid,W_grid)
        
        x1 = F.relu(self.BN1(self.LIN1(input)))
        x2 = self.BN2(self.LIN2(x1))
        
        x3a = torch.cat((x2,F.relu(self.LIN3a(x2))),dim=1)
        x3b = torch.cat((x3a,F.relu(self.LIN3b(self.BN3a(x3a)))),dim=1)
        x3c = torch.cat((x3b,F.relu(self.LIN3c(self.BN3b(x3b)))),dim=1)
        x3d = torch.cat((x3c,F.relu(self.LIN3d(self.BN3c(x3c)))),dim=1)

        x4 = self.LIN4(self.BN3d(x3d))
        #return half-resolution segmentation/prediction 
        return F.interpolate(x4, size=[self.half_res[0],self.half_res[1],self.half_res[2]], mode='trilinear',align_corners=False)
Example #5
Source File: densenet.py    From video_feature_extractor with Apache License 2.0 5 votes vote down vote up 
def forward(self, x):
        features = self.features(x)
        out = F.relu(features, inplace=True)
        last_duration = math.ceil(self.sample_duration / 16)
        last_size = math.floor(self.sample_size / 32)
        out = F.avg_pool3d(out, kernel_size=(last_duration, last_size, last_size)).view(features.size(0), -1)
        if self.last_fc:
            out = self.classifier(out)
        return out
Example #6
Source File: resnext.py    From video-caption.pytorch with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(out.size(0), planes - out.size(1),
                             out.size(2), out.size(3),
                             out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #7
Source File: wide_resnet.py    From video_feature_extractor with Apache License 2.0 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(out.size(0), planes - out.size(1),
                             out.size(2), out.size(3),
                             out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #8
Source File: models.py    From moments_models with BSD 2-Clause "Simplified" License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(
        out.size(0), planes - out.size(1),
        out.size(2), out.size(3), out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()
    out = torch.cat([out.data, zero_pads], dim=1)
    return out
Example #9
Source File: resnext.py    From GAN_Review with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(
        out.size(0), planes - out.size(1), out.size(2), out.size(3),
        out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #10
Source File: resnext_3d.py    From GAN_Review with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(
        out.size(0), planes - out.size(1), out.size(2), out.size(3),
        out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #11
Source File: wide_resnet_3d.py    From GAN_Review with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(
        out.size(0), planes - out.size(1), out.size(2), out.size(3),
        out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #12
Source File: pre_act_resnet_3d.py    From GAN_Review with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(
        out.size(0), planes - out.size(1), out.size(2), out.size(3),
        out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #13
Source File: ResNet3DMedNet.py    From MedicalZooPytorch with MIT License 5 votes vote down vote up 
def _downsample_basic_block(self, x, planes, stride):
        out = F.avg_pool3d(x, kernel_size=1, stride=stride)
        zero_pads = torch.zeros(out.size(0), planes - out.size(1), out.size(2),
                                out.size(3), out.size(4))
        if isinstance(out.data, torch.cuda.FloatTensor):
            zero_pads = zero_pads.cuda()

        out = torch.cat([out.data, zero_pads], dim=1)

        return out
Example #14
Source File: resnet.py    From ClipShots_basline with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(out.size(0), planes - out.size(1),
                             out.size(2), out.size(3),
                             out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #15
Source File: test_pyprof_nvtx.py    From apex with BSD 3-Clause "New" or "Revised" License 5 votes vote down vote up 
def test_avg_pool3d(self):
        inp = torch.randn(1, 3, 16, 224, 224, device='cuda', dtype=self.dtype)
        out = F.avg_pool3d(inp, kernel_size=5, stride=2, padding=2, ceil_mode=True, count_include_pad=False)
Example #16
Source File: DenseNet3D.py    From T3D with MIT License 5 votes vote down vote up 
def forward(self, x):
        features = self.features(x)
        out = F.relu(features, inplace=True)
        out = F.avg_pool3d(out, kernel_size=(1,7,7)).view(features.size(0), -1)
        out = self.classifier(out)
        return out
Example #17
Source File: stn.py    From istn with Apache License 2.0 5 votes vote down vote up 
def forward(self, x):
        xs = F.avg_pool3d(F.relu(self.conv1(x)), 2)
        xs = F.avg_pool3d(F.relu(self.conv2(xs)), 2)
        xs = F.avg_pool3d(F.relu(self.conv3(xs)), 2)
        xs = xs.view(xs.size(0), -1)
        xs = F.relu(self.fc(xs))
        # theta = self.affine_regressor(xs).view(-1, 3, 4)
        self.theta = self.affine_matrix(xs)

        # extract first channel for warping
        img = x.narrow(dim=1, start=0, length=1)

        # warp image
        return self.warp_image(img).to(self.device)
Example #18
Source File: resnet.py    From video_feature_extractor with Apache License 2.0 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(out.size(0), planes - out.size(1),
                             out.size(2), out.size(3),
                             out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #19
Source File: resnext.py    From video_feature_extractor with Apache License 2.0 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(out.size(0), planes - out.size(1),
                             out.size(2), out.size(3),
                             out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #20
Source File: pre_act_resnet.py    From SDN with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(
        out.size(0), planes - out.size(1), out.size(2), out.size(3),
        out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #21
Source File: wide_resnet.py    From SDN with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(
        out.size(0), planes - out.size(1), out.size(2), out.size(3),
        out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #22
Source File: densenet.py    From SDN with MIT License 5 votes vote down vote up 
def forward(self, x):
        features = self.features(x)
        out = F.relu(features, inplace=True)
        last_duration = int(math.ceil(self.sample_duration / 16))
        last_size = int(math.floor(self.sample_size / 32))
        out = F.avg_pool3d(
            out, kernel_size=(last_duration, last_size, last_size)).view(
                features.size(0), -1)
        out = self.classifier(out)
        return out
Example #23
Source File: resnet.py    From SDN with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(
        out.size(0), planes - out.size(1), out.size(2), out.size(3),
        out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #24
Source File: resnext.py    From SDN with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(
        out.size(0), planes - out.size(1), out.size(2), out.size(3),
        out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #25
Source File: resnet.py    From Efficient-3DCNNs with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(
        out.size(0), planes - out.size(1), out.size(2), out.size(3),
        out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #26
Source File: mobilenet.py    From Efficient-3DCNNs with MIT License 5 votes vote down vote up 
def forward(self, x):
        x = self.features(x)
        x = F.avg_pool3d(x, x.data.size()[-3:])
        x = x.view(x.size(0), -1)
        x = self.classifier(x)
        return x
Example #27
Source File: shufflenetv2.py    From Efficient-3DCNNs with MIT License 5 votes vote down vote up 
def forward(self, x):
        out = self.conv1(x)
        out = self.maxpool(out)
        out = self.features(out)
        out = self.conv_last(out)
        out = F.avg_pool3d(out, out.data.size()[-3:])
        out = out.view(out.size(0), -1)
        out = self.classifier(out)
        return out
Example #28
Source File: mobilenetv2.py    From Efficient-3DCNNs with MIT License 5 votes vote down vote up 
def forward(self, x):
        x = self.features(x)
        x = F.avg_pool3d(x, x.data.size()[-3:])
        x = x.view(x.size(0), -1)
        x = self.classifier(x)
        return x
Example #29
Source File: resnext.py    From Efficient-3DCNNs with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(
        out.size(0), planes - out.size(1), out.size(2), out.size(3),
        out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out
Example #30
Source File: resnet.py    From GAN_Review with MIT License 5 votes vote down vote up 
def downsample_basic_block(x, planes, stride):
    out = F.avg_pool3d(x, kernel_size=1, stride=stride)
    zero_pads = torch.Tensor(
        out.size(0), planes - out.size(1), out.size(2), out.size(3),
        out.size(4)).zero_()
    if isinstance(out.data, torch.cuda.FloatTensor):
        zero_pads = zero_pads.cuda()

    out = Variable(torch.cat([out.data, zero_pads], dim=1))

    return out