nets/yolo.py

import torch
import torch.nn as nn
from torch.nn import functional as F
from nets.CSPdarknet_tiny import darknet_tiny


class BasicConv(nn.Module):
    def __init__(self, in_channels, out_channels, kernel_size, stride=1):
        super(BasicConv, self).__init__()

        self.conv = nn.Conv2d(in_channels, out_channels, kernel_size, stride, kernel_size//2, bias=False)
        self.bn = nn.BatchNorm2d(out_channels)
        self.activation = nn.LeakyReLU(0.1)

    def forward(self, x):
        x = self.conv(x)
        x = self.bn(x)
        x = self.activation(x)
        return x

class Upsample(nn.Module):
    def __init__(self, in_channels, out_channels):
        super(Upsample, self).__init__()

        self.upsample = nn.Sequential(
            BasicConv(in_channels, out_channels, 1),
            nn.Upsample(scale_factor=2, mode='nearest')
        )

    def forward(self, x,):
        x = self.upsample(x)
        return x

def yolo_head(filters_list, in_filters):
    m = nn.Sequential(
        BasicConv(in_filters, filters_list[0], 3),
        nn.Conv2d(filters_list[0], filters_list[1], 1),
    )
    return m

class ConvBNReLU(nn.Module):
    '''Module for the Conv-BN-ReLU tuple.'''

    def __init__(self, c_in, c_out, kernel_size, stride, padding, dilation,
                 use_relu=True):
        super(ConvBNReLU, self).__init__()
        self.conv = nn.Conv2d(
            c_in, c_out, kernel_size=kernel_size, stride=stride,
            padding=padding, dilation=dilation, bias=False)
        self.bn = nn.BatchNorm2d(c_out)
        if use_relu:
            self.relu = nn.ReLU(inplace=True)
        else:
            self.relu = None

    def forward(self, x):
        x = self.conv(x)
        x = self.bn(x)
        if self.relu is not None:
            x = self.relu(x)
        return x


class CARAFE(nn.Module):
    def __init__(self, c, c_mid=64, scale=2, k_up=5, k_enc=3):
        """ The unofficial implementation of the CARAFE module.

        The details are in "https://arxiv.org/abs/1905.02188".

        Args:
            c: The channel number of the input and the output.
            c_mid: The channel number after compression.
            scale: The expected upsample scale.
            k_up: The size of the reassembly kernel.
            k_enc: The kernel size of the encoder.

        Returns:
            X: The upsampled feature map.
        """
        super(CARAFE, self).__init__()
        self.scale = scale

        self.comp = ConvBNReLU(c, c_mid, kernel_size=1, stride=1,
                               padding=0, dilation=1)
        self.enc = ConvBNReLU(c_mid, (scale * k_up) ** 2, kernel_size=k_enc,
                              stride=1, padding=k_enc // 2, dilation=1,
                              use_relu=False)
        self.pix_shf = nn.PixelShuffle(scale)

        self.upsmp = nn.Upsample(scale_factor=scale, mode='nearest')
        self.unfold = nn.Unfold(kernel_size=k_up, dilation=scale,
                                padding=k_up // 2 * scale)

    def forward(self, X):
        b, c, h, w = X.size()
        h_, w_ = h * self.scale, w * self.scale

        W = self.comp(X)  # b * m * h * w
        W = self.enc(W)  # b * 100 * h * w
        W = self.pix_shf(W)  # b * 25 * h_ * w_
        W = F.softmax(W, dim=1)  # b * 25 * h_ * w_

        X = self.upsmp(X)  # b * c * h_ * w_
        X = self.unfold(X)  # b * 25c * h_ * w_
        X = X.view(b, c, -1, h_, w_)  # b * 25 * c * h_ * w_

        X = torch.einsum('bkhw,bckhw->bchw', [W, X])  # b * c * h_ * w_
        return X

#---------------------------------------------------#
#   yolo_body--MSFNet
#---------------------------------------------------#
class YoloBody(nn.Module):
    def __init__(self, anchors_mask, num_classes, phi=0, backbone ='tiny', pretrained=False):
        super(YoloBody, self).__init__()
        if backbone == 'tiny':
            self.backbone  = darknet_tiny(pretrained)
            self.conv_for_P5    = BasicConv(512,256,1)
            self.yolo_headP5    = yolo_head([512, len(anchors_mask[0]) * (5 + num_classes)],256)
            self.upsample_1       = Upsample(256,128)
            self.conv1 = BasicConv(256,128,1)
            self.upsample_2 = CARAFE(128)
            self.yolo_headP4    = yolo_head([256, len(anchors_mask[1]) * (5 + num_classes)],384)

    def forward(self, x):

        feat1, feat2 = self.backbone(x)
        # 13,13,512 -> 13,13,256
        P5 = self.conv_for_P5(feat2)
        # 13,13,256 -> 13,13,512 -> 13,13,255
        out0 = self.yolo_headP5(P5)

        P6 = self.conv_for_P5(feat2)
        P6_Upsample = self.upsample_1(P6)

        # 13,13,256 -> 13,13,128 -> 26,26,128  
        P5 = self.conv1(P5)
        P5_Upsample = self.upsample_2(P5)

        sum = P5_Upsample + P6_Upsample
        # 26,26,256 + 26,26,128 -> 26,26,384
        # if 1 <= self.phi and self.phi <= 4:
        #     P5_Upsample = self.upsample_att(P5_Upsample)
        # P4 = torch.cat([P5_Upsample, feat1],axis=1)
        P4 = torch.cat([sum, feat1],axis=1)
        # 26,26,384 -> 26,26,256 -> 26,26,255
        out1 = self.yolo_headP4(P4)
        
        return out0, out1