NEU-DET-with-yolov8

The dataset I am using is NEU-DET, which uses yolov8 and its improved models (including Coordinate Attention and Swin Transformer) for defect detection.

1.Environmental configuration

Follow the official instructions of YOLOv8 for environment configuration.(This needs to be set as your own path)

cd cd /autodl-tmp/ultralytics-main/
pip install -r requirements.txt

2. datasets

The dataset is located in the NEU-DET folder in the data folder.

data.yaml

train: /root/autodl-tmp/ultralytics-main/data/NEU-DET/train # The directory corresponding to the training set val: /root/autodl-tmp/ultralytics-main/data/NEU-DET/test # The directory corresponding to the testing set nc: 6 # Number of tags names: ['crazing', 'inclusion', 'patches', 'pitted_surface', 'rolled-in_scale', 'scratches'] # tag name

3.xml_to_txt.py

The original dataset is in xml files format, but yolov8 requires txt files.

The purpose of this py file is to convert .xml files into .txt files.

The data in the data folder has already been converted.

4.yolov8.yaml

Making the following changes:

4. train_final.py

This is a run file.

from ultralytics import YOLO

# Load a model
# model = YOLO('yolov8s.yaml')  # build a new model from YAML
model = YOLO('/autodl-tmp/ultralytics-main/yolov8n.pt')  # load a pretrained model (recommended for training)
# model = YOLO('yolov8s.yaml').load('yolov8s.pt')  # build from YAML and transfer weights

# Train the model
if __name__ == '__main__':
    model.train(data='/autodl-tmp/ultralytics-main/data/data.yaml', epochs=200, imgsz=640, device='0')

You can execute the following command in the terminal to run it: (This needs to be set as your own path)

cd /autodl-tmp/ultralytics-main/
python train_final.py

This way, the YOLOv8 algorithm can be used for object detection.

For more detailed explanations, please refer to my article:

https://zhuanlan.zhihu.com/p/666040746

---

Next, we will use yolov8+Swin Transformer for object detection.

1.backbone

In the following path: ultralytics/cfg/models/v8/ add

"yolov8_one_swinTrans.yaml" (This is a one layer structure)

"yolov8_three_swinTrans.yaml" (This is a three layers structure)

2.conv.py

In the following path: ultralytics/nn/modules/conv.py add the following codes:

class WindowAttention(nn.Module):

    def __init__(self, dim, window_size, num_heads, qkv_bias=True, qk_scale=None, attn_drop=0., proj_drop=0.):

        super().__init__()
        self.dim = dim
        self.window_size = window_size  # Wh, Ww
        self.num_heads = num_heads
        head_dim = dim // num_heads
        self.scale = qk_scale or head_dim ** -0.5

        # define a parameter table of relative position bias
        self.relative_position_bias_table = nn.Parameter(
            torch.zeros((2 * window_size[0] - 1) * (2 * window_size[1] - 1), num_heads))  # 2*Wh-1 * 2*Ww-1, nH

        # get pair-wise relative position index for each token inside the window
        coords_h = torch.arange(self.window_size[0])
        coords_w = torch.arange(self.window_size[1])
        coords = torch.stack(torch.meshgrid([coords_h, coords_w]))  # 2, Wh, Ww
        coords_flatten = torch.flatten(coords, 1)  # 2, Wh*Ww
        relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :]  # 2, Wh*Ww, Wh*Ww
        relative_coords = relative_coords.permute(1, 2, 0).contiguous()  # Wh*Ww, Wh*Ww, 2
        relative_coords[:, :, 0] += self.window_size[0] - 1  # shift to start from 0
        relative_coords[:, :, 1] += self.window_size[1] - 1
        relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1
        relative_position_index = relative_coords.sum(-1)  # Wh*Ww, Wh*Ww
        self.register_buffer("relative_position_index", relative_position_index)

        self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)
        self.attn_drop = nn.Dropout(attn_drop)
        self.proj = nn.Linear(dim, dim)
        self.proj_drop = nn.Dropout(proj_drop)

        nn.init.normal_(self.relative_position_bias_table, std=.02)
        self.softmax = nn.Softmax(dim=-1)

    def forward(self, x, mask=None):

        B_, N, C = x.shape
        qkv = self.qkv(x).reshape(B_, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
        q, k, v = qkv[0], qkv[1], qkv[2]  # make torchscript happy (cannot use tensor as tuple)

        q = q * self.scale
        attn = (q @ k.transpose(-2, -1))

        relative_position_bias = self.relative_position_bias_table[self.relative_position_index.view(-1)].view(
            self.window_size[0] * self.window_size[1], self.window_size[0] * self.window_size[1], -1)  # Wh*Ww,Wh*Ww,nH
        relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous()  # nH, Wh*Ww, Wh*Ww
        attn = attn + relative_position_bias.unsqueeze(0)

        if mask is not None:
            nW = mask.shape[0]
            attn = attn.view(B_ // nW, nW, self.num_heads, N, N) + mask.unsqueeze(1).unsqueeze(0)
            attn = attn.view(-1, self.num_heads, N, N)
            attn = self.softmax(attn)
        else:
            attn = self.softmax(attn)

        attn = self.attn_drop(attn)

        # print(attn.dtype, v.dtype)
        try:
            x = (attn @ v).transpose(1, 2).reshape(B_, N, C)
        except:
            # print(attn.dtype, v.dtype)
            x = (attn.half() @ v).transpose(1, 2).reshape(B_, N, C)
        x = self.proj(x)
        x = self.proj_drop(x)
        return x


class SwinTransformer(nn.Module):
    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
        super(SwinTransformer, self).__init__()
        c_ = int(c2 * e)  # hidden channels
        self.cv1 = Conv(c1, c_, 1, 1)
        self.cv2 = Conv(c1, c_, 1, 1)
        self.cv3 = Conv(2 * c_, c2, 1, 1)
        num_heads = c_ // 32
        self.m = SwinTransformerBlock(c_, c_, num_heads, n)
        # self.m = nn.Sequential(*[Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])

    def forward(self, x):
        y1 = self.m(self.cv1(x))
        y2 = self.cv2(x)
        return self.cv3(torch.cat((y1, y2), dim=1))


class SwinTransformerB(nn.Module):
    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
    def __init__(self, c1, c2, n=1, shortcut=False, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
        super(Swin_Transformer_B, self).__init__()
        c_ = int(c2)  # hidden channels
        self.cv1 = Conv(c1, c_, 1, 1)
        self.cv2 = Conv(c_, c_, 1, 1)
        self.cv3 = Conv(2 * c_, c2, 1, 1)
        num_heads = c_ // 32
        self.m = SwinTransformerBlock(c_, c_, num_heads, n)
        # self.m = nn.Sequential(*[Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])

    def forward(self, x):
        x1 = self.cv1(x)
        y1 = self.m(x1)
        y2 = self.cv2(x1)
        return self.cv3(torch.cat((y1, y2), dim=1))


class SwinTransformerC(nn.Module):
    # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):  # ch_in, ch_out, number, shortcut, groups, expansion
        super(Swin_Transformer_C, self).__init__()
        c_ = int(c2 * e)  # hidden channels
        self.cv1 = Conv(c1, c_, 1, 1)
        self.cv2 = Conv(c1, c_, 1, 1)
        self.cv3 = Conv(c_, c_, 1, 1)
        self.cv4 = Conv(2 * c_, c2, 1, 1)
        num_heads = c_ // 32
        self.m = SwinTransformerBlock(c_, c_, num_heads, n)
        # self.m = nn.Sequential(*[Bottleneck(c_, c_, shortcut, g, e=1.0) for _ in range(n)])

    def forward(self, x):
        y1 = self.cv3(self.m(self.cv1(x)))
        y2 = self.cv2(x)
        return self.cv4(torch.cat((y1, y2), dim=1))


class Mlp(nn.Module):

    def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.SiLU, drop=0.):
        super().__init__()
        out_features = out_features or in_features
        hidden_features = hidden_features or in_features
        self.fc1 = nn.Linear(in_features, hidden_features)
        self.act = act_layer()
        self.fc2 = nn.Linear(hidden_features, out_features)
        self.drop = nn.Dropout(drop)

    def forward(self, x):
        x = self.fc1(x)
        x = self.act(x)
        x = self.drop(x)
        x = self.fc2(x)
        x = self.drop(x)
        return x


def window_partition(x, window_size):
    B, H, W, C = x.shape
    assert H % window_size == 0, 'feature map h and w can not divide by window size'
    x = x.view(B, H // window_size, window_size, W // window_size, window_size, C)
    windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C)
    return windows


def window_reverse(windows, window_size, H, W):
    B = int(windows.shape[0] / (H * W / window_size / window_size))
    x = windows.view(B, H // window_size, W // window_size, window_size, window_size, -1)
    x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1)
    return x


class SwinTransformerLayer(nn.Module):

    def __init__(self, dim, num_heads, window_size=8, shift_size=0,
                 mlp_ratio=4., qkv_bias=True, qk_scale=None, drop=0., attn_drop=0., drop_path=0.,
                 act_layer=nn.SiLU, norm_layer=nn.LayerNorm):
        super().__init__()
        self.dim = dim
        self.num_heads = num_heads
        self.window_size = window_size
        self.shift_size = shift_size
        self.mlp_ratio = mlp_ratio
        # if min(self.input_resolution) <= self.window_size:
        #     # if window size is larger than input resolution, we don't partition windows