core/update_noctx.py

import torch
import torch.nn as nn
import torch.nn.init as init
import torch.nn.functional as F
from opt_einsum import contract

class DispHead(nn.Module):
    def __init__(self, input_dim=128, hidden_dim=256, output_dim=1):
        super(DispHead, self).__init__()
        self.conv1 = nn.Conv2d(input_dim, hidden_dim, 3, padding=1)
        self.conv2 = nn.Conv2d(hidden_dim, output_dim, 3, padding=1)
        self.relu = nn.ReLU(inplace=True)

    def forward(self, x):
        return self.conv2(self.relu(self.conv1(x)))

class ConvGRU_NoCTX(nn.Module):
    def __init__(self, hidden_dim, input_dim, kernel_size=3):
        super(ConvGRU_NoCTX, self).__init__()
        self.convz = nn.Conv2d(hidden_dim+input_dim, hidden_dim, kernel_size, padding=kernel_size//2)
        self.convr = nn.Conv2d(hidden_dim+input_dim, hidden_dim, kernel_size, padding=kernel_size//2)
        self.convq = nn.Conv2d(hidden_dim+input_dim, hidden_dim, kernel_size, padding=kernel_size//2)

        self._initialize_weights()

    def forward(self, h, *x_list):
        x = torch.cat(x_list, dim=1)
        hx = torch.cat([h, x], dim=1)

        z = torch.sigmoid(self.convz(hx) )
        r = torch.sigmoid(self.convr(hx) )
        q = torch.tanh(self.convq(torch.cat([r*h, x], dim=1)) )

        h = (1-z) * h + z * q
        return h
    
    def _initialize_weights(self):
        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                init.kaiming_normal_(m.weight)
                if m.bias is not None:
                    m.bias.data.zero_()

class SepConvGRU(nn.Module):
    def __init__(self, hidden_dim=128, input_dim=192+128):
        super(SepConvGRU, self).__init__()
        self.convz1 = nn.Conv2d(hidden_dim+input_dim, hidden_dim, (1,5), padding=(0,2))
        self.convr1 = nn.Conv2d(hidden_dim+input_dim, hidden_dim, (1,5), padding=(0,2))
        self.convq1 = nn.Conv2d(hidden_dim+input_dim, hidden_dim, (1,5), padding=(0,2))

        self.convz2 = nn.Conv2d(hidden_dim+input_dim, hidden_dim, (5,1), padding=(2,0))
        self.convr2 = nn.Conv2d(hidden_dim+input_dim, hidden_dim, (5,1), padding=(2,0))
        self.convq2 = nn.Conv2d(hidden_dim+input_dim, hidden_dim, (5,1), padding=(2,0))


    def forward(self, h, *x):
        # horizontal
        x = torch.cat(x, dim=1)
        hx = torch.cat([h, x], dim=1)
        z = torch.sigmoid(self.convz1(hx))
        r = torch.sigmoid(self.convr1(hx))
        q = torch.tanh(self.convq1(torch.cat([r*h, x], dim=1)))        
        h = (1-z) * h + z * q

        # vertical
        hx = torch.cat([h, x], dim=1)
        z = torch.sigmoid(self.convz2(hx))
        r = torch.sigmoid(self.convr2(hx))
        q = torch.tanh(self.convq2(torch.cat([r*h, x], dim=1)))       
        h = (1-z) * h + z * q

        return h

class BasicShiftEncoder(nn.Module):
    def __init__(self, args):
        super(BasicShiftEncoder, self).__init__()
        self.args = args

        cor_planes = args.corr_levels * (2*args.corr_radius + 1)

        self.convc1 = nn.Conv2d(cor_planes, 64, 1, padding=0)
        self.convc2 = nn.Conv2d(64, 64, 3, padding=1)
        self.convf1 = nn.Conv2d(1, 64, 7, padding=3)
        self.convf2 = nn.Conv2d(64, 64, 3, padding=1)
        self.conv = nn.Conv2d(64+64, 128-1, 3, padding=1)

    def forward(self, disp, corr):
        cor = F.relu(self.convc1(corr))
        cor = F.relu(self.convc2(cor))
        dis = F.relu(self.convf1(disp))
        dis = F.relu(self.convf2(dis))

        cor_dis = torch.cat([cor, dis], dim=1)
        out = F.relu(self.conv(cor_dis))
        return torch.cat([out, disp], dim=1)

def pool2x(x):
    return F.avg_pool2d(x, 3, stride=2, padding=1)

def pool4x(x):
    return F.avg_pool2d(x, 5, stride=4, padding=1)

def interp(x, dest):
    interp_args = {'mode': 'bilinear', 'align_corners': True}
    return F.interpolate(x, dest.shape[2:], **interp_args)


class DispBasicMultiUpdateBlock_NoCTX(nn.Module):
    def __init__(self, args, hidden_dims=[]):
        super(DispBasicMultiUpdateBlock_NoCTX, self).__init__()
        self.args = args
        self.encoder = BasicShiftEncoder(args)
        encoder_output_dim = 128

        self.gru08 = ConvGRU_NoCTX(hidden_dims[2], encoder_output_dim + hidden_dims[1] * (args.n_gru_layers > 1))
        self.gru16 = ConvGRU_NoCTX(hidden_dims[1], hidden_dims[0] * (args.n_gru_layers == 3) + hidden_dims[2])
        self.gru32 = ConvGRU_NoCTX(hidden_dims[0], hidden_dims[1])
        self.disp_head = DispHead(hidden_dims[2], hidden_dim=256, output_dim=1)
        factor = 2**self.args.n_downsample

        self.mask = nn.Sequential(
            nn.Conv2d(hidden_dims[2], 256, 3, padding=1),
            nn.ReLU(inplace=True),
            nn.Conv2d(256, (factor**2)*9, 1, padding=0))

    def forward(self, net, corr=None, disp=None, iter08=True, iter16=True, iter32=True, update=True):

        if iter32:
            net[2] = self.gru32(net[2], pool2x(net[1]))
        if iter16:
            if self.args.n_gru_layers > 2:
                net[1] = self.gru16(net[1], pool2x(net[0]), interp(net[2], net[1]))
            else:
                net[1] = self.gru16(net[1], pool2x(net[0]))
        if iter08:
            motion_features = self.encoder(disp, corr)
            if self.args.n_gru_layers > 1:
                net[0] = self.gru08(net[0], motion_features, interp(net[1], net[0]))
            else:
                net[0] = self.gru08(net[0], motion_features)

        if not update:
            return net

        delta_disp = self.disp_head(net[0])

        # scale mask to balence gradients
        mask = .25 * self.mask(net[0])
        return net, mask, delta_disp