انتقل إلى المحتوى

مرجع ل ultralytics/models/sam/modules/tiny_encoder.py

ملاحظه

هذا الملف متاح في https://github.com/ultralytics/ultralytics/ نقطة / الرئيسية /ultralytics/نماذج/sam/الوحدات/tiny_encoder.py. إذا اكتشفت مشكلة ، فيرجى المساعدة في إصلاحها من خلال المساهمة في طلب 🛠️ سحب. شكرا لك 🙏!



ultralytics.models.sam.modules.tiny_encoder.Conv2d_BN

قواعد: Sequential

حاوية متسلسلة تؤدي التفاف 2D متبوعا بتطبيع الدفعات.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
class Conv2d_BN(torch.nn.Sequential):
    """A sequential container that performs 2D convolution followed by batch normalization."""

    def __init__(self, a, b, ks=1, stride=1, pad=0, dilation=1, groups=1, bn_weight_init=1):
        """Initializes the MBConv model with given input channels, output channels, expansion ratio, activation, and
        drop path.
        """
        super().__init__()
        self.add_module("c", torch.nn.Conv2d(a, b, ks, stride, pad, dilation, groups, bias=False))
        bn = torch.nn.BatchNorm2d(b)
        torch.nn.init.constant_(bn.weight, bn_weight_init)
        torch.nn.init.constant_(bn.bias, 0)
        self.add_module("bn", bn)

__init__(a, b, ks=1, stride=1, pad=0, dilation=1, groups=1, bn_weight_init=1)

تهيئة نموذج MBConv مع قنوات إدخال معينة وقنوات الإخراج ونسبة التوسع والتنشيط و إسقاط المسار.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
def __init__(self, a, b, ks=1, stride=1, pad=0, dilation=1, groups=1, bn_weight_init=1):
    """Initializes the MBConv model with given input channels, output channels, expansion ratio, activation, and
    drop path.
    """
    super().__init__()
    self.add_module("c", torch.nn.Conv2d(a, b, ks, stride, pad, dilation, groups, bias=False))
    bn = torch.nn.BatchNorm2d(b)
    torch.nn.init.constant_(bn.weight, bn_weight_init)
    torch.nn.init.constant_(bn.bias, 0)
    self.add_module("bn", bn)



ultralytics.models.sam.modules.tiny_encoder.PatchEmbed

قواعد: Module

يدمج الصور في تصحيحات ويعرضها في بعد تضمين محدد.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
38 39 40 41 42 43 44 45 46 47 48 49 50 51525354555657 585960
class PatchEmbed(nn.Module):
    """Embeds images into patches and projects them into a specified embedding dimension."""

    def __init__(self, in_chans, embed_dim, resolution, activation):
        """Initialize the PatchMerging class with specified input, output dimensions, resolution and activation
        function.
        """
        super().__init__()
        img_size: Tuple[int, int] = to_2tuple(resolution)
        self.patches_resolution = (img_size[0] // 4, img_size[1] // 4)
        self.num_patches = self.patches_resolution[0] * self.patches_resolution[1]
        self.in_chans = in_chans
        self.embed_dim = embed_dim
        n = embed_dim
        self.seq = nn.Sequential(
            Conv2d_BN(in_chans, n // 2, 3, 2, 1),
            activation(),
            Conv2d_BN(n // 2, n, 3, 2, 1),
        )

    def forward(self, x):
        """Runs input tensor 'x' through the PatchMerging model's sequence of operations."""
        return self.seq(x)

__init__(in_chans, embed_dim, resolution, activation)

تهيئة فئة PatchMerging مع الإدخال المحدد وأبعاد الإخراج والدقة والتنشيط دالة.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
41 42 43 44 45 46 47484950 5152535455 56
def __init__(self, in_chans, embed_dim, resolution, activation):
    """Initialize the PatchMerging class with specified input, output dimensions, resolution and activation
    function.
    """
    super().__init__()
    img_size: Tuple[int, int] = to_2tuple(resolution)
    self.patches_resolution = (img_size[0] // 4, img_size[1] // 4)
    self.num_patches = self.patches_resolution[0] * self.patches_resolution[1]
    self.in_chans = in_chans
    self.embed_dim = embed_dim
    n = embed_dim
    self.seq = nn.Sequential(
        Conv2d_BN(in_chans, n // 2, 3, 2, 1),
        activation(),
        Conv2d_BN(n // 2, n, 3, 2, 1),
    )

forward(x)

تشغيل الإدخال tensor "x" من خلال تسلسل عمليات نموذج PatchMerging.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
def forward(self, x):
    """Runs input tensor 'x' through the PatchMerging model's sequence of operations."""
    return self.seq(x)



ultralytics.models.sam.modules.tiny_encoder.MBConv

قواعد: Module

طبقة عنق الزجاجة المقلوبة المتنقلة (MBConv) ، وهي جزء من بنية EfficientNet.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 7879 8081 82  83 84 85 86 87 888990 9192 93 9495969798
class MBConv(nn.Module):
    """Mobile Inverted Bottleneck Conv (MBConv) layer, part of the EfficientNet architecture."""

    def __init__(self, in_chans, out_chans, expand_ratio, activation, drop_path):
        """Initializes a convolutional layer with specified dimensions, input resolution, depth, and activation
        function.
        """
        super().__init__()
        self.in_chans = in_chans
        self.hidden_chans = int(in_chans * expand_ratio)
        self.out_chans = out_chans

        self.conv1 = Conv2d_BN(in_chans, self.hidden_chans, ks=1)
        self.act1 = activation()

        self.conv2 = Conv2d_BN(self.hidden_chans, self.hidden_chans, ks=3, stride=1, pad=1, groups=self.hidden_chans)
        self.act2 = activation()

        self.conv3 = Conv2d_BN(self.hidden_chans, out_chans, ks=1, bn_weight_init=0.0)
        self.act3 = activation()

        # NOTE: `DropPath` is needed only for training.
        # self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
        self.drop_path = nn.Identity()

    def forward(self, x):
        """Implements the forward pass for the model architecture."""
        shortcut = x
        x = self.conv1(x)
        x = self.act1(x)
        x = self.conv2(x)
        x = self.act2(x)
        x = self.conv3(x)
        x = self.drop_path(x)
        x += shortcut
        return self.act3(x)

__init__(in_chans, out_chans, expand_ratio, activation, drop_path)

تهيئة طبقة تلافيفية بأبعاد محددة، دقة الإدخال، العمق، والتنشيط دالة.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
66 67 68 69 70 71 72 73 74 75 76 777879808182838485 86
def __init__(self, in_chans, out_chans, expand_ratio, activation, drop_path):
    """Initializes a convolutional layer with specified dimensions, input resolution, depth, and activation
    function.
    """
    super().__init__()
    self.in_chans = in_chans
    self.hidden_chans = int(in_chans * expand_ratio)
    self.out_chans = out_chans

    self.conv1 = Conv2d_BN(in_chans, self.hidden_chans, ks=1)
    self.act1 = activation()

    self.conv2 = Conv2d_BN(self.hidden_chans, self.hidden_chans, ks=3, stride=1, pad=1, groups=self.hidden_chans)
    self.act2 = activation()

    self.conv3 = Conv2d_BN(self.hidden_chans, out_chans, ks=1, bn_weight_init=0.0)
    self.act3 = activation()

    # NOTE: `DropPath` is needed only for training.
    # self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
    self.drop_path = nn.Identity()

forward(x)

ينفذ التمريرة الأمامية لبنية النموذج.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
def forward(self, x):
    """Implements the forward pass for the model architecture."""
    shortcut = x
    x = self.conv1(x)
    x = self.act1(x)
    x = self.conv2(x)
    x = self.act2(x)
    x = self.conv3(x)
    x = self.drop_path(x)
    x += shortcut
    return self.act3(x)



ultralytics.models.sam.modules.tiny_encoder.PatchMerging

قواعد: Module

يدمج التصحيحات المجاورة في خريطة المعالم والمشاريع إلى بعد جديد.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127128129 130 131132 133
class PatchMerging(nn.Module):
    """Merges neighboring patches in the feature map and projects to a new dimension."""

    def __init__(self, input_resolution, dim, out_dim, activation):
        """Initializes the ConvLayer with specific dimension, input resolution, depth, activation, drop path, and other
        optional parameters.
        """
        super().__init__()

        self.input_resolution = input_resolution
        self.dim = dim
        self.out_dim = out_dim
        self.act = activation()
        self.conv1 = Conv2d_BN(dim, out_dim, 1, 1, 0)
        stride_c = 1 if out_dim in [320, 448, 576] else 2
        self.conv2 = Conv2d_BN(out_dim, out_dim, 3, stride_c, 1, groups=out_dim)
        self.conv3 = Conv2d_BN(out_dim, out_dim, 1, 1, 0)

    def forward(self, x):
        """Applies forward pass on the input utilizing convolution and activation layers, and returns the result."""
        if x.ndim == 3:
            H, W = self.input_resolution
            B = len(x)
            # (B, C, H, W)
            x = x.view(B, H, W, -1).permute(0, 3, 1, 2)

        x = self.conv1(x)
        x = self.act(x)

        x = self.conv2(x)
        x = self.act(x)
        x = self.conv3(x)
        return x.flatten(2).transpose(1, 2)

__init__(input_resolution, dim, out_dim, activation)

تهيئة ConvLayer ببعد محدد ودقة الإدخال والعمق والتنشيط ومسار الإسقاط وغيرها المعلمات الاختيارية.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
104 105 106 107 108 109 110 111 112 113 114 115116 117
def __init__(self, input_resolution, dim, out_dim, activation):
    """Initializes the ConvLayer with specific dimension, input resolution, depth, activation, drop path, and other
    optional parameters.
    """
    super().__init__()

    self.input_resolution = input_resolution
    self.dim = dim
    self.out_dim = out_dim
    self.act = activation()
    self.conv1 = Conv2d_BN(dim, out_dim, 1, 1, 0)
    stride_c = 1 if out_dim in [320, 448, 576] else 2
    self.conv2 = Conv2d_BN(out_dim, out_dim, 3, stride_c, 1, groups=out_dim)
    self.conv3 = Conv2d_BN(out_dim, out_dim, 1, 1, 0)

forward(x)

يطبق تمريرة أمامية على الإدخال باستخدام طبقات الالتفاف والتنشيط، ويعيد النتيجة.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
119 120 121 122 123 124 125 126 127128129 130 131 132133
def forward(self, x):
    """Applies forward pass on the input utilizing convolution and activation layers, and returns the result."""
    if x.ndim == 3:
        H, W = self.input_resolution
        B = len(x)
        # (B, C, H, W)
        x = x.view(B, H, W, -1).permute(0, 3, 1, 2)

    x = self.conv1(x)
    x = self.act(x)

    x = self.conv2(x)
    x = self.act(x)
    x = self.conv3(x)
    return x.flatten(2).transpose(1, 2)



ultralytics.models.sam.modules.tiny_encoder.ConvLayer

قواعد: Module

طبقة تلافيفية تتميز بتلافيف عنق الزجاجة المقلوبة المتعددة على غرار MobileNetV3 (MBConv).

يطبق اختياريا عمليات الاختزال على المخرجات، ويوفر الدعم لنقطة فحص التدرج.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158159 160 161 162 163 164 165 166167168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187188189 190191 192 193 194 195 196197 198199200
class ConvLayer(nn.Module):
    """
    Convolutional Layer featuring multiple MobileNetV3-style inverted bottleneck convolutions (MBConv).

    Optionally applies downsample operations to the output, and provides support for gradient checkpointing.
    """

    def __init__(
        self,
        dim,
        input_resolution,
        depth,
        activation,
        drop_path=0.0,
        downsample=None,
        use_checkpoint=False,
        out_dim=None,
        conv_expand_ratio=4.0,
    ):
        """
        Initializes the ConvLayer with the given dimensions and settings.

        Args:
            dim (int): The dimensionality of the input and output.
            input_resolution (Tuple[int, int]): The resolution of the input image.
            depth (int): The number of MBConv layers in the block.
            activation (Callable): Activation function applied after each convolution.
            drop_path (Union[float, List[float]]): Drop path rate. Single float or a list of floats for each MBConv.
            downsample (Optional[Callable]): Function for downsampling the output. None to skip downsampling.
            use_checkpoint (bool): Whether to use gradient checkpointing to save memory.
            out_dim (Optional[int]): The dimensionality of the output. None means it will be the same as `dim`.
            conv_expand_ratio (float): Expansion ratio for the MBConv layers.
        """
        super().__init__()
        self.dim = dim
        self.input_resolution = input_resolution
        self.depth = depth
        self.use_checkpoint = use_checkpoint

        # Build blocks
        self.blocks = nn.ModuleList(
            [
                MBConv(
                    dim,
                    dim,
                    conv_expand_ratio,
                    activation,
                    drop_path[i] if isinstance(drop_path, list) else drop_path,
                )
                for i in range(depth)
            ]
        )

        # Patch merging layer
        self.downsample = (
            None
            if downsample is None
            else downsample(input_resolution, dim=dim, out_dim=out_dim, activation=activation)
        )

    def forward(self, x):
        """Processes the input through a series of convolutional layers and returns the activated output."""
        for blk in self.blocks:
            x = checkpoint.checkpoint(blk, x) if self.use_checkpoint else blk(x)
        return x if self.downsample is None else self.downsample(x)

__init__(dim, input_resolution, depth, activation, drop_path=0.0, downsample=None, use_checkpoint=False, out_dim=None, conv_expand_ratio=4.0)

تهيئة ConvLayer بالأبعاد والإعدادات المحددة.

البارامترات:

اسم نوع وصف افتراضي
dim int

أبعاد المدخلات والمخرجات.

مطلوب
input_resolution Tuple[int, int]

دقة صورة الإدخال.

مطلوب
depth int

عدد طبقات MBConv في الكتلة.

مطلوب
activation Callable

يتم تطبيق وظيفة التنشيط بعد كل التفاف.

مطلوب
drop_path Union[float, List[float]]

انخفاض معدل المسار. تعويم واحد أو قائمة عوامات لكل MBConv.

0.0
downsample Optional[Callable]

وظيفة لاختزال الإخراج. لا شيء لتخطي الاختزال.

None
use_checkpoint bool

ما إذا كنت تريد استخدام نقطة التحقق المتدرجة لتوفير الذاكرة.

False
out_dim Optional[int]

أبعاد الإخراج. لا شيء يعني أنه سيكون هو نفسه dim.

None
conv_expand_ratio float

نسبة التوسع لطبقات MBConv.

4.0
شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167168169 170171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188189 190191 192193 194
def __init__(
    self,
    dim,
    input_resolution,
    depth,
    activation,
    drop_path=0.0,
    downsample=None,
    use_checkpoint=False,
    out_dim=None,
    conv_expand_ratio=4.0,
):
    """
    Initializes the ConvLayer with the given dimensions and settings.

    Args:
        dim (int): The dimensionality of the input and output.
        input_resolution (Tuple[int, int]): The resolution of the input image.
        depth (int): The number of MBConv layers in the block.
        activation (Callable): Activation function applied after each convolution.
        drop_path (Union[float, List[float]]): Drop path rate. Single float or a list of floats for each MBConv.
        downsample (Optional[Callable]): Function for downsampling the output. None to skip downsampling.
        use_checkpoint (bool): Whether to use gradient checkpointing to save memory.
        out_dim (Optional[int]): The dimensionality of the output. None means it will be the same as `dim`.
        conv_expand_ratio (float): Expansion ratio for the MBConv layers.
    """
    super().__init__()
    self.dim = dim
    self.input_resolution = input_resolution
    self.depth = depth
    self.use_checkpoint = use_checkpoint

    # Build blocks
    self.blocks = nn.ModuleList(
        [
            MBConv(
                dim,
                dim,
                conv_expand_ratio,
                activation,
                drop_path[i] if isinstance(drop_path, list) else drop_path,
            )
            for i in range(depth)
        ]
    )

    # Patch merging layer
    self.downsample = (
        None
        if downsample is None
        else downsample(input_resolution, dim=dim, out_dim=out_dim, activation=activation)
    )

forward(x)

يعالج الإدخال من خلال سلسلة من الطبقات التلافيفية ويعيد المخرجات المنشطة.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
196 197 198199200 
def forward(self, x):
    """Processes the input through a series of convolutional layers and returns the activated output."""
    for blk in self.blocks:
        x = checkpoint.checkpoint(blk, x) if self.use_checkpoint else blk(x)
    return x if self.downsample is None else self.downsample(x)



ultralytics.models.sam.modules.tiny_encoder.Mlp

قواعد: Module

متعدد الطبقات Perceptron (MLP) لمعماريات المحولات.

تأخذ هذه الطبقة مدخلا مع in_features ، وتطبق تطبيع الطبقة وطبقتين متصلتين بالكامل.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218219 220221222 223 224 225 226 227228
class Mlp(nn.Module):
    """
    Multi-layer Perceptron (MLP) for transformer architectures.

    This layer takes an input with in_features, applies layer normalization and two fully-connected layers.
    """

    def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.0):
        """Initializes Attention module with the given parameters including dimension, key_dim, number of heads, etc."""
        super().__init__()
        out_features = out_features or in_features
        hidden_features = hidden_features or in_features
        self.norm = nn.LayerNorm(in_features)
        self.fc1 = nn.Linear(in_features, hidden_features)
        self.fc2 = nn.Linear(hidden_features, out_features)
        self.act = act_layer()
        self.drop = nn.Dropout(drop)

    def forward(self, x):
        """Applies operations on input x and returns modified x, runs downsample if not None."""
        x = self.norm(x)
        x = self.fc1(x)
        x = self.act(x)
        x = self.drop(x)
        x = self.fc2(x)
        return self.drop(x)

__init__(in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.0)

تهيئة وحدة الانتباه مع المعلمات المحددة بما في ذلك الأبعاد ، key_dim ، عدد الرؤوس ، إلخ.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
210 211 212 213 214 215 216 217 218 219
def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.0):
    """Initializes Attention module with the given parameters including dimension, key_dim, number of heads, etc."""
    super().__init__()
    out_features = out_features or in_features
    hidden_features = hidden_features or in_features
    self.norm = nn.LayerNorm(in_features)
    self.fc1 = nn.Linear(in_features, hidden_features)
    self.fc2 = nn.Linear(hidden_features, out_features)
    self.act = act_layer()
    self.drop = nn.Dropout(drop)

forward(x)

يطبق العمليات على الإدخال x ويقوم بإرجاع x المعدل ، ويقوم بتشغيل العينة السفلية إن لم يكن لا شيء.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
221 222 223 224 225 226 227 228
def forward(self, x):
    """Applies operations on input x and returns modified x, runs downsample if not None."""
    x = self.norm(x)
    x = self.fc1(x)
    x = self.act(x)
    x = self.drop(x)
    x = self.fc2(x)
    return self.drop(x)



ultralytics.models.sam.modules.tiny_encoder.Attention

قواعد: Module

وحدة انتباه متعددة الرؤوس مع دعم الوعي المكاني ، وتطبيق تحيزات الانتباه على أساس المكاني دقة. ينفذ تحيزات الانتباه القابلة للتدريب لكل إزاحة فريدة بين المواضع المكانية في الدقة شبكة.

سمات:

اسم نوع وصف
ab Tensor

تحيزات الانتباه المخزنة مؤقتا للاستدلال ، يتم حذفها أثناء التدريب.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248249 250 251 252 253 254 255256257258 259 260 261262263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285286 287288289290 291292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318319320321
class Attention(torch.nn.Module):
    """
    Multi-head attention module with support for spatial awareness, applying attention biases based on spatial
    resolution. Implements trainable attention biases for each unique offset between spatial positions in the resolution
    grid.

    Attributes:
        ab (Tensor, optional): Cached attention biases for inference, deleted during training.
    """

    def __init__(
        self,
        dim,
        key_dim,
        num_heads=8,
        attn_ratio=4,
        resolution=(14, 14),
    ):
        """
        Initializes the Attention module.

        Args:
            dim (int): The dimensionality of the input and output.
            key_dim (int): The dimensionality of the keys and queries.
            num_heads (int, optional): Number of attention heads. Default is 8.
            attn_ratio (float, optional): Attention ratio, affecting the dimensions of the value vectors. Default is 4.
            resolution (Tuple[int, int], optional): Spatial resolution of the input feature map. Default is (14, 14).

        Raises:
            AssertionError: If `resolution` is not a tuple of length 2.
        """
        super().__init__()

        assert isinstance(resolution, tuple) and len(resolution) == 2
        self.num_heads = num_heads
        self.scale = key_dim**-0.5
        self.key_dim = key_dim
        self.nh_kd = nh_kd = key_dim * num_heads
        self.d = int(attn_ratio * key_dim)
        self.dh = int(attn_ratio * key_dim) * num_heads
        self.attn_ratio = attn_ratio
        h = self.dh + nh_kd * 2

        self.norm = nn.LayerNorm(dim)
        self.qkv = nn.Linear(dim, h)
        self.proj = nn.Linear(self.dh, dim)

        points = list(itertools.product(range(resolution[0]), range(resolution[1])))
        N = len(points)
        attention_offsets = {}
        idxs = []
        for p1 in points:
            for p2 in points:
                offset = (abs(p1[0] - p2[0]), abs(p1[1] - p2[1]))
                if offset not in attention_offsets:
                    attention_offsets[offset] = len(attention_offsets)
                idxs.append(attention_offsets[offset])
        self.attention_biases = torch.nn.Parameter(torch.zeros(num_heads, len(attention_offsets)))
        self.register_buffer("attention_bias_idxs", torch.LongTensor(idxs).view(N, N), persistent=False)

    @torch.no_grad()
    def train(self, mode=True):
        """Sets the module in training mode and handles attribute 'ab' based on the mode."""
        super().train(mode)
        if mode and hasattr(self, "ab"):
            del self.ab
        else:
            self.ab = self.attention_biases[:, self.attention_bias_idxs]

    def forward(self, x):  # x
        """Performs forward pass over the input tensor 'x' by applying normalization and querying keys/values."""
        B, N, _ = x.shape  # B, N, C

        # Normalization
        x = self.norm(x)

        qkv = self.qkv(x)
        # (B, N, num_heads, d)
        q, k, v = qkv.view(B, N, self.num_heads, -1).split([self.key_dim, self.key_dim, self.d], dim=3)
        # (B, num_heads, N, d)
        q = q.permute(0, 2, 1, 3)
        k = k.permute(0, 2, 1, 3)
        v = v.permute(0, 2, 1, 3)
        self.ab = self.ab.to(self.attention_biases.device)

        attn = (q @ k.transpose(-2, -1)) * self.scale + (
            self.attention_biases[:, self.attention_bias_idxs] if self.training else self.ab
        )
        attn = attn.softmax(dim=-1)
        x = (attn @ v).transpose(1, 2).reshape(B, N, self.dh)
        return self.proj(x)

__init__(dim, key_dim, num_heads=8, attn_ratio=4, resolution=(14, 14))

تهيئة وحدة الانتباه.

البارامترات:

اسم نوع وصف افتراضي
dim int

أبعاد المدخلات والمخرجات.

مطلوب
key_dim int

أبعاد المفاتيح والاستعلامات.

مطلوب
num_heads int

عدد رؤوس الانتباه. الافتراضي هو 8.

8
attn_ratio float

نسبة الانتباه ، مما يؤثر على أبعاد متجهات القيمة. الافتراضي هو 4.

4
resolution Tuple[int, int]

الدقة المكانية لخريطة معلم الإدخال. الافتراضي هو (14 ، 14).

(14, 14)

يثير:

نوع وصف
AssertionError

لو resolution ليست مجموعة طولها 2.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259260 261 262 263 264 265 266 267 268269 270 271272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287288 289
def __init__(
    self,
    dim,
    key_dim,
    num_heads=8,
    attn_ratio=4,
    resolution=(14, 14),
):
    """
    Initializes the Attention module.

    Args:
        dim (int): The dimensionality of the input and output.
        key_dim (int): The dimensionality of the keys and queries.
        num_heads (int, optional): Number of attention heads. Default is 8.
        attn_ratio (float, optional): Attention ratio, affecting the dimensions of the value vectors. Default is 4.
        resolution (Tuple[int, int], optional): Spatial resolution of the input feature map. Default is (14, 14).

    Raises:
        AssertionError: If `resolution` is not a tuple of length 2.
    """
    super().__init__()

    assert isinstance(resolution, tuple) and len(resolution) == 2
    self.num_heads = num_heads
    self.scale = key_dim**-0.5
    self.key_dim = key_dim
    self.nh_kd = nh_kd = key_dim * num_heads
    self.d = int(attn_ratio * key_dim)
    self.dh = int(attn_ratio * key_dim) * num_heads
    self.attn_ratio = attn_ratio
    h = self.dh + nh_kd * 2

    self.norm = nn.LayerNorm(dim)
    self.qkv = nn.Linear(dim, h)
    self.proj = nn.Linear(self.dh, dim)

    points = list(itertools.product(range(resolution[0]), range(resolution[1])))
    N = len(points)
    attention_offsets = {}
    idxs = []
    for p1 in points:
        for p2 in points:
            offset = (abs(p1[0] - p2[0]), abs(p1[1] - p2[1]))
            if offset not in attention_offsets:
                attention_offsets[offset] = len(attention_offsets)
            idxs.append(attention_offsets[offset])
    self.attention_biases = torch.nn.Parameter(torch.zeros(num_heads, len(attention_offsets)))
    self.register_buffer("attention_bias_idxs", torch.LongTensor(idxs).view(N, N), persistent=False)

forward(x)

ينفذ تمريرا أماميا فوق الإدخال tensor "x" من خلال تطبيق التطبيع والاستعلام عن المفاتيح / القيم.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
300 301 302 303 304 305 306 307 308309 310 311 312 313 314 315 316 317 318319 320 321
def forward(self, x):  # x
    """Performs forward pass over the input tensor 'x' by applying normalization and querying keys/values."""
    B, N, _ = x.shape  # B, N, C

    # Normalization
    x = self.norm(x)

    qkv = self.qkv(x)
    # (B, N, num_heads, d)
    q, k, v = qkv.view(B, N, self.num_heads, -1).split([self.key_dim, self.key_dim, self.d], dim=3)
    # (B, num_heads, N, d)
    q = q.permute(0, 2, 1, 3)
    k = k.permute(0, 2, 1, 3)
    v = v.permute(0, 2, 1, 3)
    self.ab = self.ab.to(self.attention_biases.device)

    attn = (q @ k.transpose(-2, -1)) * self.scale + (
        self.attention_biases[:, self.attention_bias_idxs] if self.training else self.ab
    )
    attn = attn.softmax(dim=-1)
    x = (attn @ v).transpose(1, 2).reshape(B, N, self.dh)
    return self.proj(x)

train(mode=True)

يضبط الوحدة في وضع التدريب ويعالج السمة "ab" بناء على الوضع.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
291 292 293 294 295 296 297 298
@torch.no_grad()
def train(self, mode=True):
    """Sets the module in training mode and handles attribute 'ab' based on the mode."""
    super().train(mode)
    if mode and hasattr(self, "ab"):
        del self.ab
    else:
        self.ab = self.attention_biases[:, self.attention_bias_idxs]



ultralytics.models.sam.modules.tiny_encoder.TinyViTBlock

قواعد: Module

TinyViT Block الذي يطبق الاهتمام الذاتي والتفاف محلي على المدخلات.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350351 352 353 354355 356 357 358 359 360 361 362 363 364 365 366 367 368369 370 371 372 373 374 375 376 377 378379 380381 382 383 384385386 387 388 389 390 391 392 393 394 395 396 397 398399 400 401 402 403 404 405406 407 408409 410411412413414 415 416 417 418 419 420 421 422 423 424 425 426 427 428429 430431432433 434
class TinyViTBlock(nn.Module):
    """TinyViT Block that applies self-attention and a local convolution to the input."""

    def __init__(
        self,
        dim,
        input_resolution,
        num_heads,
        window_size=7,
        mlp_ratio=4.0,
        drop=0.0,
        drop_path=0.0,
        local_conv_size=3,
        activation=nn.GELU,
    ):
        """
        Initializes the TinyViTBlock.

        Args:
            dim (int): The dimensionality of the input and output.
            input_resolution (Tuple[int, int]): Spatial resolution of the input feature map.
            num_heads (int): Number of attention heads.
            window_size (int, optional): Window size for attention. Default is 7.
            mlp_ratio (float, optional): Ratio of mlp hidden dim to embedding dim. Default is 4.
            drop (float, optional): Dropout rate. Default is 0.
            drop_path (float, optional): Stochastic depth rate. Default is 0.
            local_conv_size (int, optional): The kernel size of the local convolution. Default is 3.
            activation (torch.nn, optional): Activation function for MLP. Default is nn.GELU.

        Raises:
            AssertionError: If `window_size` is not greater than 0.
            AssertionError: If `dim` is not divisible by `num_heads`.
        """
        super().__init__()
        self.dim = dim
        self.input_resolution = input_resolution
        self.num_heads = num_heads
        assert window_size > 0, "window_size must be greater than 0"
        self.window_size = window_size
        self.mlp_ratio = mlp_ratio

        # NOTE: `DropPath` is needed only for training.
        # self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
        self.drop_path = nn.Identity()

        assert dim % num_heads == 0, "dim must be divisible by num_heads"
        head_dim = dim // num_heads

        window_resolution = (window_size, window_size)
        self.attn = Attention(dim, head_dim, num_heads, attn_ratio=1, resolution=window_resolution)

        mlp_hidden_dim = int(dim * mlp_ratio)
        mlp_activation = activation
        self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=mlp_activation, drop=drop)

        pad = local_conv_size // 2
        self.local_conv = Conv2d_BN(dim, dim, ks=local_conv_size, stride=1, pad=pad, groups=dim)

    def forward(self, x):
        """Applies attention-based transformation or padding to input 'x' before passing it through a local
        convolution.
        """
        H, W = self.input_resolution
        B, L, C = x.shape
        assert L == H * W, "input feature has wrong size"
        res_x = x
        if H == self.window_size and W == self.window_size:
            x = self.attn(x)
        else:
            x = x.view(B, H, W, C)
            pad_b = (self.window_size - H % self.window_size) % self.window_size
            pad_r = (self.window_size - W % self.window_size) % self.window_size
            padding = pad_b > 0 or pad_r > 0

            if padding:
                x = F.pad(x, (0, 0, 0, pad_r, 0, pad_b))

            pH, pW = H + pad_b, W + pad_r
            nH = pH // self.window_size
            nW = pW // self.window_size
            # Window partition
            x = (
                x.view(B, nH, self.window_size, nW, self.window_size, C)
                .transpose(2, 3)
                .reshape(B * nH * nW, self.window_size * self.window_size, C)
            )
            x = self.attn(x)
            # Window reverse
            x = x.view(B, nH, nW, self.window_size, self.window_size, C).transpose(2, 3).reshape(B, pH, pW, C)

            if padding:
                x = x[:, :H, :W].contiguous()

            x = x.view(B, L, C)

        x = res_x + self.drop_path(x)

        x = x.transpose(1, 2).reshape(B, C, H, W)
        x = self.local_conv(x)
        x = x.view(B, C, L).transpose(1, 2)

        return x + self.drop_path(self.mlp(x))

    def extra_repr(self) -> str:
        """Returns a formatted string representing the TinyViTBlock's parameters: dimension, input resolution, number of
        attentions heads, window size, and MLP ratio.
        """
        return (
            f"dim={self.dim}, input_resolution={self.input_resolution}, num_heads={self.num_heads}, "
            f"window_size={self.window_size}, mlp_ratio={self.mlp_ratio}"
        )

__init__(dim, input_resolution, num_heads, window_size=7, mlp_ratio=4.0, drop=0.0, drop_path=0.0, local_conv_size=3, activation=nn.GELU)

تهيئة TinyViTBlock.

البارامترات:

اسم نوع وصف افتراضي
dim int

أبعاد المدخلات والمخرجات.

مطلوب
input_resolution Tuple[int, int]

الدقة المكانية لخريطة معلم الإدخال.

مطلوب
num_heads int

عدد رؤوس الانتباه.

مطلوب
window_size int

حجم النافذة للانتباه. الافتراضي هو 7.

7
mlp_ratio float

نسبة mlp المخفية خافتة إلى تضمين قاتمة. الافتراضي هو 4.

4.0
drop float

معدل التسرب. الافتراضي هو 0.

0.0
drop_path float

معدل عمق مؤشر ستوكاستيك. الافتراضي هو 0.

0.0
local_conv_size int

حجم نواة الالتفاف المحلي. الافتراضي هو 3.

3
activation nn

وظيفة التنشيط ل MLP. الافتراضي هو nn. جيلو.

GELU

يثير:

نوع وصف
AssertionError

لو window_size ليس أكبر من 0.

AssertionError

لو dim لا يقبل القسمة على num_heads.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
327 328 329 330 331 332 333 334 335 336 337 338339 340 341 342 343 344 345 346 347 348 349 350 351 352 353354355 356357358359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375376377 378 379 380
def __init__(
    self,
    dim,
    input_resolution,
    num_heads,
    window_size=7,
    mlp_ratio=4.0,
    drop=0.0,
    drop_path=0.0,
    local_conv_size=3,
    activation=nn.GELU,
):
    """
    Initializes the TinyViTBlock.

    Args:
        dim (int): The dimensionality of the input and output.
        input_resolution (Tuple[int, int]): Spatial resolution of the input feature map.
        num_heads (int): Number of attention heads.
        window_size (int, optional): Window size for attention. Default is 7.
        mlp_ratio (float, optional): Ratio of mlp hidden dim to embedding dim. Default is 4.
        drop (float, optional): Dropout rate. Default is 0.
        drop_path (float, optional): Stochastic depth rate. Default is 0.
        local_conv_size (int, optional): The kernel size of the local convolution. Default is 3.
        activation (torch.nn, optional): Activation function for MLP. Default is nn.GELU.

    Raises:
        AssertionError: If `window_size` is not greater than 0.
        AssertionError: If `dim` is not divisible by `num_heads`.
    """
    super().__init__()
    self.dim = dim
    self.input_resolution = input_resolution
    self.num_heads = num_heads
    assert window_size > 0, "window_size must be greater than 0"
    self.window_size = window_size
    self.mlp_ratio = mlp_ratio

    # NOTE: `DropPath` is needed only for training.
    # self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
    self.drop_path = nn.Identity()

    assert dim % num_heads == 0, "dim must be divisible by num_heads"
    head_dim = dim // num_heads

    window_resolution = (window_size, window_size)
    self.attn = Attention(dim, head_dim, num_heads, attn_ratio=1, resolution=window_resolution)

    mlp_hidden_dim = int(dim * mlp_ratio)
    mlp_activation = activation
    self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=mlp_activation, drop=drop)

    pad = local_conv_size // 2
    self.local_conv = Conv2d_BN(dim, dim, ks=local_conv_size, stride=1, pad=pad, groups=dim)

extra_repr()

إرجاع سلسلة منسقة تمثل معلمات TinyViTBlock: البعد ودقة الإدخال وعدد رؤوس الانتباه وحجم النافذة ونسبة MLP.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
427428 429 430 431 432 433434
def extra_repr(self) -> str:
    """Returns a formatted string representing the TinyViTBlock's parameters: dimension, input resolution, number of
    attentions heads, window size, and MLP ratio.
    """
    return (
        f"dim={self.dim}, input_resolution={self.input_resolution}, num_heads={self.num_heads}, "
        f"window_size={self.window_size}, mlp_ratio={self.mlp_ratio}"
    )

forward(x)

يطبق التحويل أو الحشو المستند إلى الانتباه على الإدخال "x" قبل تمريره عبر محلي الالتواء.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397398 399 400 401 402 403 404 405 406407 408409410 411412413414 415 416 417 418 419 420 421 422 423 424425
def forward(self, x):
    """Applies attention-based transformation or padding to input 'x' before passing it through a local
    convolution.
    """
    H, W = self.input_resolution
    B, L, C = x.shape
    assert L == H * W, "input feature has wrong size"
    res_x = x
    if H == self.window_size and W == self.window_size:
        x = self.attn(x)
    else:
        x = x.view(B, H, W, C)
        pad_b = (self.window_size - H % self.window_size) % self.window_size
        pad_r = (self.window_size - W % self.window_size) % self.window_size
        padding = pad_b > 0 or pad_r > 0

        if padding:
            x = F.pad(x, (0, 0, 0, pad_r, 0, pad_b))

        pH, pW = H + pad_b, W + pad_r
        nH = pH // self.window_size
        nW = pW // self.window_size
        # Window partition
        x = (
            x.view(B, nH, self.window_size, nW, self.window_size, C)
            .transpose(2, 3)
            .reshape(B * nH * nW, self.window_size * self.window_size, C)
        )
        x = self.attn(x)
        # Window reverse
        x = x.view(B, nH, nW, self.window_size, self.window_size, C).transpose(2, 3).reshape(B, pH, pW, C)

        if padding:
            x = x[:, :H, :W].contiguous()

        x = x.view(B, L, C)

    x = res_x + self.drop_path(x)

    x = x.transpose(1, 2).reshape(B, C, H, W)
    x = self.local_conv(x)
    x = x.view(B, C, L).transpose(1, 2)

    return x + self.drop_path(self.mlp(x))



ultralytics.models.sam.modules.tiny_encoder.BasicLayer

قواعد: Module

طبقة TinyViT أساسية لمرحلة واحدة في بنية TinyViT.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
437 438 439 440 441 442 443 444 445 446 447 448449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464465 466 467468469 470 471 472 473 474 475 476 477 478479 480 481 482 483 484 485 486 487 488489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508509 510 511 512 513 514515 516
class BasicLayer(nn.Module):
    """A basic TinyViT layer for one stage in a TinyViT architecture."""

    def __init__(
        self,
        dim,
        input_resolution,
        depth,
        num_heads,
        window_size,
        mlp_ratio=4.0,
        drop=0.0,
        drop_path=0.0,
        downsample=None,
        use_checkpoint=False,
        local_conv_size=3,
        activation=nn.GELU,
        out_dim=None,
    ):
        """
        Initializes the BasicLayer.

        Args:
            dim (int): The dimensionality of the input and output.
            input_resolution (Tuple[int, int]): Spatial resolution of the input feature map.
            depth (int): Number of TinyViT blocks.
            num_heads (int): Number of attention heads.
            window_size (int): Local window size.
            mlp_ratio (float, optional): Ratio of mlp hidden dim to embedding dim. Default is 4.
            drop (float, optional): Dropout rate. Default is 0.
            drop_path (float | tuple[float], optional): Stochastic depth rate. Default is 0.
            downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default is None.
            use_checkpoint (bool, optional): Whether to use checkpointing to save memory. Default is False.
            local_conv_size (int, optional): Kernel size of the local convolution. Default is 3.
            activation (torch.nn, optional): Activation function for MLP. Default is nn.GELU.
            out_dim (int | None, optional): The output dimension of the layer. Default is None.

        Raises:
            ValueError: If `drop_path` is a list of float but its length doesn't match `depth`.
        """
        super().__init__()
        self.dim = dim
        self.input_resolution = input_resolution
        self.depth = depth
        self.use_checkpoint = use_checkpoint

        # Build blocks
        self.blocks = nn.ModuleList(
            [
                TinyViTBlock(
                    dim=dim,
                    input_resolution=input_resolution,
                    num_heads=num_heads,
                    window_size=window_size,
                    mlp_ratio=mlp_ratio,
                    drop=drop,
                    drop_path=drop_path[i] if isinstance(drop_path, list) else drop_path,
                    local_conv_size=local_conv_size,
                    activation=activation,
                )
                for i in range(depth)
            ]
        )

        # Patch merging layer
        self.downsample = (
            None
            if downsample is None
            else downsample(input_resolution, dim=dim, out_dim=out_dim, activation=activation)
        )

    def forward(self, x):
        """Performs forward propagation on the input tensor and returns a normalized tensor."""
        for blk in self.blocks:
            x = checkpoint.checkpoint(blk, x) if self.use_checkpoint else blk(x)
        return x if self.downsample is None else self.downsample(x)

    def extra_repr(self) -> str:
        """Returns a string representation of the extra_repr function with the layer's parameters."""
        return f"dim={self.dim}, input_resolution={self.input_resolution}, depth={self.depth}"

__init__(dim, input_resolution, depth, num_heads, window_size, mlp_ratio=4.0, drop=0.0, drop_path=0.0, downsample=None, use_checkpoint=False, local_conv_size=3, activation=nn.GELU, out_dim=None)

تهيئة الطبقة الأساسية.

البارامترات:

اسم نوع وصف افتراضي
dim int

أبعاد المدخلات والمخرجات.

مطلوب
input_resolution Tuple[int, int]

الدقة المكانية لخريطة معلم الإدخال.

مطلوب
depth int

عدد كتل TinyViT.

مطلوب
num_heads int

عدد رؤوس الانتباه.

مطلوب
window_size int

حجم النافذة المحلية.

مطلوب
mlp_ratio float

نسبة mlp المخفية خافتة إلى تضمين قاتمة. الافتراضي هو 4.

4.0
drop float

معدل التسرب. الافتراضي هو 0.

0.0
drop_path float | tuple[float]

معدل عمق مؤشر ستوكاستيك. الافتراضي هو 0.

0.0
downsample Module | None

طبقة العينة المصغرة في نهاية الطبقة. الافتراضي هو بلا.

None
use_checkpoint bool

ما إذا كنت تريد استخدام Checkpoint لحفظ الذاكرة. الافتراضي هو خطأ.

False
local_conv_size int

حجم نواة الالتفاف المحلي. الافتراضي هو 3.

3
activation nn

وظيفة التنشيط ل MLP. الافتراضي هو nn. جيلو.

GELU
out_dim int | None

بعد الإخراج للطبقة. الافتراضي هو بلا.

None

يثير:

نوع وصف
ValueError

لو drop_path هي قائمة عائمة ولكن طولها غير متطابق depth.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457458459 460 461 462 463 464 465 466467 468469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485486 487 488 489 490 491 492 493 494 495 496 497 498 499 500501502 503 504 505 506
def __init__(
    self,
    dim,
    input_resolution,
    depth,
    num_heads,
    window_size,
    mlp_ratio=4.0,
    drop=0.0,
    drop_path=0.0,
    downsample=None,
    use_checkpoint=False,
    local_conv_size=3,
    activation=nn.GELU,
    out_dim=None,
):
    """
    Initializes the BasicLayer.

    Args:
        dim (int): The dimensionality of the input and output.
        input_resolution (Tuple[int, int]): Spatial resolution of the input feature map.
        depth (int): Number of TinyViT blocks.
        num_heads (int): Number of attention heads.
        window_size (int): Local window size.
        mlp_ratio (float, optional): Ratio of mlp hidden dim to embedding dim. Default is 4.
        drop (float, optional): Dropout rate. Default is 0.
        drop_path (float | tuple[float], optional): Stochastic depth rate. Default is 0.
        downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default is None.
        use_checkpoint (bool, optional): Whether to use checkpointing to save memory. Default is False.
        local_conv_size (int, optional): Kernel size of the local convolution. Default is 3.
        activation (torch.nn, optional): Activation function for MLP. Default is nn.GELU.
        out_dim (int | None, optional): The output dimension of the layer. Default is None.

    Raises:
        ValueError: If `drop_path` is a list of float but its length doesn't match `depth`.
    """
    super().__init__()
    self.dim = dim
    self.input_resolution = input_resolution
    self.depth = depth
    self.use_checkpoint = use_checkpoint

    # Build blocks
    self.blocks = nn.ModuleList(
        [
            TinyViTBlock(
                dim=dim,
                input_resolution=input_resolution,
                num_heads=num_heads,
                window_size=window_size,
                mlp_ratio=mlp_ratio,
                drop=drop,
                drop_path=drop_path[i] if isinstance(drop_path, list) else drop_path,
                local_conv_size=local_conv_size,
                activation=activation,
            )
            for i in range(depth)
        ]
    )

    # Patch merging layer
    self.downsample = (
        None
        if downsample is None
        else downsample(input_resolution, dim=dim, out_dim=out_dim, activation=activation)
    )

extra_repr()

ترجع تمثيل سلسلة لدالة extra_repr بمعلمات الطبقة.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
def extra_repr(self) -> str:
    """Returns a string representation of the extra_repr function with the layer's parameters."""
    return f"dim={self.dim}, input_resolution={self.input_resolution}, depth={self.depth}"

forward(x)

ينفذ الانتشار الأمامي على الإدخال tensor وإرجاع تطبيع tensor.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
def forward(self, x):
    """Performs forward propagation on the input tensor and returns a normalized tensor."""
    for blk in self.blocks:
        x = checkpoint.checkpoint(blk, x) if self.use_checkpoint else blk(x)
    return x if self.downsample is None else self.downsample(x)



ultralytics.models.sam.modules.tiny_encoder.LayerNorm2d

قواعد: Module

A PyTorch تنفيذ تطبيع الطبقة في 2D.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
519 520 521 522 523 524 525 526 527528529 530 531 532 533534
class LayerNorm2d(nn.Module):
    """A PyTorch implementation of Layer Normalization in 2D."""

    def __init__(self, num_channels: int, eps: float = 1e-6) -> None:
        """Initialize LayerNorm2d with the number of channels and an optional epsilon."""
        super().__init__()
        self.weight = nn.Parameter(torch.ones(num_channels))
        self.bias = nn.Parameter(torch.zeros(num_channels))
        self.eps = eps

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        """Perform a forward pass, normalizing the input tensor."""
        u = x.mean(1, keepdim=True)
        s = (x - u).pow(2).mean(1, keepdim=True)
        x = (x - u) / torch.sqrt(s + self.eps)
        return self.weight[:, None, None] * x + self.bias[:, None, None]

__init__(num_channels, eps=1e-06)

قم بتهيئة LayerNorm2d بعدد القنوات وإبسيلون اختياري.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
522 523 524 525 526 527
def __init__(self, num_channels: int, eps: float = 1e-6) -> None:
    """Initialize LayerNorm2d with the number of channels and an optional epsilon."""
    super().__init__()
    self.weight = nn.Parameter(torch.ones(num_channels))
    self.bias = nn.Parameter(torch.zeros(num_channels))
    self.eps = eps

forward(x)

قم بإجراء تمريرة أمامية ، وتطبيع الإدخال tensor.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
529 530 531 532 533 534
def forward(self, x: torch.Tensor) -> torch.Tensor:
    """Perform a forward pass, normalizing the input tensor."""
    u = x.mean(1, keepdim=True)
    s = (x - u).pow(2).mean(1, keepdim=True)
    x = (x - u) / torch.sqrt(s + self.eps)
    return self.weight[:, None, None] * x + self.bias[:, None, None]



ultralytics.models.sam.modules.tiny_encoder.TinyViT

قواعد: Module

بنية TinyViT لمهام الرؤية.

سمات:

اسم نوع وصف
img_size int

حجم صورة الإدخال.

in_chans int

عدد قنوات الإدخال.

num_classes int

عدد فئات التصنيف.

embed_dims List[int]

قائمة أبعاد التضمين لكل طبقة.

depths List[int]

قائمة الأعماق لكل طبقة.

num_heads List[int]

قائمة بعدد رؤوس الانتباه لكل طبقة.

window_sizes List[int]

قائمة أحجام النوافذ لكل طبقة.

mlp_ratio float

نسبة البعد المخفي MLP إلى بعد التضمين.

drop_rate float

معدل التسرب لطبقات السقوط.

drop_path_rate float

معدل مسار السقوط لعمق مؤشر ستوكاستيك.

use_checkpoint bool

استخدم نقطة الاختيار لاستخدام الذاكرة بكفاءة.

mbconv_expand_ratio float

نسبة التوسع لطبقة MBConv.

local_conv_size int

حجم نواة الالتفاف المحلية.

layer_lr_decay float

اضمحلال معدل التعلم الطبقي.

ملاحظه

يتم تعميم هذا التنفيذ لقبول قائمة الأعماق ، رؤوس الانتباه ، تضمين الأبعاد وأحجام النوافذ ، مما يسمح لك بإنشاء ملف "كومة" من نماذج TinyViT من تكوينات مختلفة.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
class TinyViT(nn.Module):
    """
    The TinyViT architecture for vision tasks.

    Attributes:
        img_size (int): Input image size.
        in_chans (int): Number of input channels.
        num_classes (int): Number of classification classes.
        embed_dims (List[int]): List of embedding dimensions for each layer.
        depths (List[int]): List of depths for each layer.
        num_heads (List[int]): List of number of attention heads for each layer.
        window_sizes (List[int]): List of window sizes for each layer.
        mlp_ratio (float): Ratio of MLP hidden dimension to embedding dimension.
        drop_rate (float): Dropout rate for drop layers.
        drop_path_rate (float): Drop path rate for stochastic depth.
        use_checkpoint (bool): Use checkpointing for efficient memory usage.
        mbconv_expand_ratio (float): Expansion ratio for MBConv layer.
        local_conv_size (int): Local convolution kernel size.
        layer_lr_decay (float): Layer-wise learning rate decay.

    Note:
        This implementation is generalized to accept a list of depths, attention heads,
        embedding dimensions and window sizes, which allows you to create a
        "stack" of TinyViT models of varying configurations.
    """

    def __init__(
        self,
        img_size=224,
        in_chans=3,
        num_classes=1000,
        embed_dims=[96, 192, 384, 768],
        depths=[2, 2, 6, 2],
        num_heads=[3, 6, 12, 24],
        window_sizes=[7, 7, 14, 7],
        mlp_ratio=4.0,
        drop_rate=0.0,
        drop_path_rate=0.1,
        use_checkpoint=False,
        mbconv_expand_ratio=4.0,
        local_conv_size=3,
        layer_lr_decay=1.0,
    ):
        """
        Initializes the TinyViT model.

        Args:
            img_size (int, optional): The input image size. Defaults to 224.
            in_chans (int, optional): Number of input channels. Defaults to 3.
            num_classes (int, optional): Number of classification classes. Defaults to 1000.
            embed_dims (List[int], optional): List of embedding dimensions for each layer. Defaults to [96, 192, 384, 768].
            depths (List[int], optional): List of depths for each layer. Defaults to [2, 2, 6, 2].
            num_heads (List[int], optional): List of number of attention heads for each layer. Defaults to [3, 6, 12, 24].
            window_sizes (List[int], optional): List of window sizes for each layer. Defaults to [7, 7, 14, 7].
            mlp_ratio (float, optional): Ratio of MLP hidden dimension to embedding dimension. Defaults to 4.
            drop_rate (float, optional): Dropout rate. Defaults to 0.
            drop_path_rate (float, optional): Drop path rate for stochastic depth. Defaults to 0.1.
            use_checkpoint (bool, optional): Whether to use checkpointing for efficient memory usage. Defaults to False.
            mbconv_expand_ratio (float, optional): Expansion ratio for MBConv layer. Defaults to 4.0.
            local_conv_size (int, optional): Local convolution kernel size. Defaults to 3.
            layer_lr_decay (float, optional): Layer-wise learning rate decay. Defaults to 1.0.
        """
        super().__init__()
        self.img_size = img_size
        self.num_classes = num_classes
        self.depths = depths
        self.num_layers = len(depths)
        self.mlp_ratio = mlp_ratio

        activation = nn.GELU

        self.patch_embed = PatchEmbed(
            in_chans=in_chans, embed_dim=embed_dims[0], resolution=img_size, activation=activation
        )

        patches_resolution = self.patch_embed.patches_resolution
        self.patches_resolution = patches_resolution

        # Stochastic depth
        dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))]  # stochastic depth decay rule

        # Build layers
        self.layers = nn.ModuleList()
        for i_layer in range(self.num_layers):
            kwargs = dict(
                dim=embed_dims[i_layer],
                input_resolution=(
                    patches_resolution[0] // (2 ** (i_layer - 1 if i_layer == 3 else i_layer)),
                    patches_resolution[1] // (2 ** (i_layer - 1 if i_layer == 3 else i_layer)),
                ),
                #   input_resolution=(patches_resolution[0] // (2 ** i_layer),
                #                     patches_resolution[1] // (2 ** i_layer)),
                depth=depths[i_layer],
                drop_path=dpr[sum(depths[:i_layer]) : sum(depths[: i_layer + 1])],
                downsample=PatchMerging if (i_layer < self.num_layers - 1) else None,
                use_checkpoint=use_checkpoint,
                out_dim=embed_dims[min(i_layer + 1, len(embed_dims) - 1)],
                activation=activation,
            )
            if i_layer == 0:
                layer = ConvLayer(conv_expand_ratio=mbconv_expand_ratio, **kwargs)
            else:
                layer = BasicLayer(
                    num_heads=num_heads[i_layer],
                    window_size=window_sizes[i_layer],
                    mlp_ratio=self.mlp_ratio,
                    drop=drop_rate,
                    local_conv_size=local_conv_size,
                    **kwargs,
                )
            self.layers.append(layer)

        # Classifier head
        self.norm_head = nn.LayerNorm(embed_dims[-1])
        self.head = nn.Linear(embed_dims[-1], num_classes) if num_classes > 0 else torch.nn.Identity()

        # Init weights
        self.apply(self._init_weights)
        self.set_layer_lr_decay(layer_lr_decay)
        self.neck = nn.Sequential(
            nn.Conv2d(
                embed_dims[-1],
                256,
                kernel_size=1,
                bias=False,
            ),
            LayerNorm2d(256),
            nn.Conv2d(
                256,
                256,
                kernel_size=3,
                padding=1,
                bias=False,
            ),
            LayerNorm2d(256),
        )

    def set_layer_lr_decay(self, layer_lr_decay):
        """Sets the learning rate decay for each layer in the TinyViT model."""
        decay_rate = layer_lr_decay

        # Layers -> blocks (depth)
        depth = sum(self.depths)
        lr_scales = [decay_rate ** (depth - i - 1) for i in range(depth)]

        def _set_lr_scale(m, scale):
            """Sets the learning rate scale for each layer in the model based on the layer's depth."""
            for p in m.parameters():
                p.lr_scale = scale

        self.patch_embed.apply(lambda x: _set_lr_scale(x, lr_scales[0]))
        i = 0
        for layer in self.layers:
            for block in layer.blocks:
                block.apply(lambda x: _set_lr_scale(x, lr_scales[i]))
                i += 1
            if layer.downsample is not None:
                layer.downsample.apply(lambda x: _set_lr_scale(x, lr_scales[i - 1]))
        assert i == depth
        for m in [self.norm_head, self.head]:
            m.apply(lambda x: _set_lr_scale(x, lr_scales[-1]))

        for k, p in self.named_parameters():
            p.param_name = k

        def _check_lr_scale(m):
            """Checks if the learning rate scale attribute is present in module's parameters."""
            for p in m.parameters():
                assert hasattr(p, "lr_scale"), p.param_name

        self.apply(_check_lr_scale)

    def _init_weights(self, m):
        """Initializes weights for linear layers and layer normalization in the given module."""
        if isinstance(m, nn.Linear):
            # NOTE: This initialization is needed only for training.
            # trunc_normal_(m.weight, std=.02)
            if m.bias is not None:
                nn.init.constant_(m.bias, 0)
        elif isinstance(m, nn.LayerNorm):
            nn.init.constant_(m.bias, 0)
            nn.init.constant_(m.weight, 1.0)

    @torch.jit.ignore
    def no_weight_decay_keywords(self):
        """Returns a dictionary of parameter names where weight decay should not be applied."""
        return {"attention_biases"}

    def forward_features(self, x):
        """Runs the input through the model layers and returns the transformed output."""
        x = self.patch_embed(x)  # x input is (N, C, H, W)

        x = self.layers[0](x)
        start_i = 1

        for i in range(start_i, len(self.layers)):
            layer = self.layers[i]
            x = layer(x)
        B, _, C = x.shape
        x = x.view(B, 64, 64, C)
        x = x.permute(0, 3, 1, 2)
        return self.neck(x)

    def forward(self, x):
        """Executes a forward pass on the input tensor through the constructed model layers."""
        return self.forward_features(x)

__init__(img_size=224, in_chans=3, num_classes=1000, embed_dims=[96, 192, 384, 768], depths=[2, 2, 6, 2], num_heads=[3, 6, 12, 24], window_sizes=[7, 7, 14, 7], mlp_ratio=4.0, drop_rate=0.0, drop_path_rate=0.1, use_checkpoint=False, mbconv_expand_ratio=4.0, local_conv_size=3, layer_lr_decay=1.0)

تهيئة نموذج TinyViT.

البارامترات:

اسم نوع وصف افتراضي
img_size int

حجم صورة الإدخال. الإعدادات الافتراضية هي 224.

224
in_chans int

عدد قنوات الإدخال. الإعدادات الافتراضية إلى 3.

3
num_classes int

عدد فئات التصنيف. الإعدادات الافتراضية إلى 1000.

1000
embed_dims List[int]

قائمة أبعاد التضمين لكل طبقة. الإعدادات الافتراضية إلى [96 ، 192 ، 384 ، 768].

[96, 192, 384, 768]
depths List[int]

قائمة الأعماق لكل طبقة. الإعدادات الافتراضية هي [2، 2، 6، 2].

[2, 2, 6, 2]
num_heads List[int]

قائمة بعدد رؤوس الانتباه لكل طبقة. الإعدادات الافتراضية هي [3، 6، 12، 24].

[3, 6, 12, 24]
window_sizes List[int]

قائمة أحجام النوافذ لكل طبقة. الإعدادات الافتراضية إلى [7، 7، 14، 7].

[7, 7, 14, 7]
mlp_ratio float

نسبة البعد المخفي MLP إلى بعد التضمين. الإعدادات الافتراضية إلى 4.

4.0
drop_rate float

معدل التسرب. الإعدادات الافتراضية هي 0.

0.0
drop_path_rate float

معدل مسار السقوط لعمق مؤشر ستوكاستيك. الإعدادات الافتراضية هي 0.1.

0.1
use_checkpoint bool

ما إذا كنت تريد استخدام نقطة التحقق لاستخدام الذاكرة بكفاءة. الإعدادات الافتراضية إلى خطأ.

False
mbconv_expand_ratio float

نسبة التوسع لطبقة MBConv. الإعدادات الافتراضية هي 4.0.

4.0
local_conv_size int

حجم نواة الالتفاف المحلية. الإعدادات الافتراضية إلى 3.

3
layer_lr_decay float

اضمحلال معدل التعلم الطبقي. الإعدادات الافتراضية هي 1.0.

1.0
شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587588589 590591 592 593594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616617618619 620621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647648649 650 651652653 654 655 656 657 658659 660 661 662 663 664 665666 667 668 669 670671672
def __init__(
    self,
    img_size=224,
    in_chans=3,
    num_classes=1000,
    embed_dims=[96, 192, 384, 768],
    depths=[2, 2, 6, 2],
    num_heads=[3, 6, 12, 24],
    window_sizes=[7, 7, 14, 7],
    mlp_ratio=4.0,
    drop_rate=0.0,
    drop_path_rate=0.1,
    use_checkpoint=False,
    mbconv_expand_ratio=4.0,
    local_conv_size=3,
    layer_lr_decay=1.0,
):
    """
    Initializes the TinyViT model.

    Args:
        img_size (int, optional): The input image size. Defaults to 224.
        in_chans (int, optional): Number of input channels. Defaults to 3.
        num_classes (int, optional): Number of classification classes. Defaults to 1000.
        embed_dims (List[int], optional): List of embedding dimensions for each layer. Defaults to [96, 192, 384, 768].
        depths (List[int], optional): List of depths for each layer. Defaults to [2, 2, 6, 2].
        num_heads (List[int], optional): List of number of attention heads for each layer. Defaults to [3, 6, 12, 24].
        window_sizes (List[int], optional): List of window sizes for each layer. Defaults to [7, 7, 14, 7].
        mlp_ratio (float, optional): Ratio of MLP hidden dimension to embedding dimension. Defaults to 4.
        drop_rate (float, optional): Dropout rate. Defaults to 0.
        drop_path_rate (float, optional): Drop path rate for stochastic depth. Defaults to 0.1.
        use_checkpoint (bool, optional): Whether to use checkpointing for efficient memory usage. Defaults to False.
        mbconv_expand_ratio (float, optional): Expansion ratio for MBConv layer. Defaults to 4.0.
        local_conv_size (int, optional): Local convolution kernel size. Defaults to 3.
        layer_lr_decay (float, optional): Layer-wise learning rate decay. Defaults to 1.0.
    """
    super().__init__()
    self.img_size = img_size
    self.num_classes = num_classes
    self.depths = depths
    self.num_layers = len(depths)
    self.mlp_ratio = mlp_ratio

    activation = nn.GELU

    self.patch_embed = PatchEmbed(
        in_chans=in_chans, embed_dim=embed_dims[0], resolution=img_size, activation=activation
    )

    patches_resolution = self.patch_embed.patches_resolution
    self.patches_resolution = patches_resolution

    # Stochastic depth
    dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))]  # stochastic depth decay rule

    # Build layers
    self.layers = nn.ModuleList()
    for i_layer in range(self.num_layers):
        kwargs = dict(
            dim=embed_dims[i_layer],
            input_resolution=(
                patches_resolution[0] // (2 ** (i_layer - 1 if i_layer == 3 else i_layer)),
                patches_resolution[1] // (2 ** (i_layer - 1 if i_layer == 3 else i_layer)),
            ),
            #   input_resolution=(patches_resolution[0] // (2 ** i_layer),
            #                     patches_resolution[1] // (2 ** i_layer)),
            depth=depths[i_layer],
            drop_path=dpr[sum(depths[:i_layer]) : sum(depths[: i_layer + 1])],
            downsample=PatchMerging if (i_layer < self.num_layers - 1) else None,
            use_checkpoint=use_checkpoint,
            out_dim=embed_dims[min(i_layer + 1, len(embed_dims) - 1)],
            activation=activation,
        )
        if i_layer == 0:
            layer = ConvLayer(conv_expand_ratio=mbconv_expand_ratio, **kwargs)
        else:
            layer = BasicLayer(
                num_heads=num_heads[i_layer],
                window_size=window_sizes[i_layer],
                mlp_ratio=self.mlp_ratio,
                drop=drop_rate,
                local_conv_size=local_conv_size,
                **kwargs,
            )
        self.layers.append(layer)

    # Classifier head
    self.norm_head = nn.LayerNorm(embed_dims[-1])
    self.head = nn.Linear(embed_dims[-1], num_classes) if num_classes > 0 else torch.nn.Identity()

    # Init weights
    self.apply(self._init_weights)
    self.set_layer_lr_decay(layer_lr_decay)
    self.neck = nn.Sequential(
        nn.Conv2d(
            embed_dims[-1],
            256,
            kernel_size=1,
            bias=False,
        ),
        LayerNorm2d(256),
        nn.Conv2d(
            256,
            256,
            kernel_size=3,
            padding=1,
            bias=False,
        ),
        LayerNorm2d(256),
    )

forward(x)

ينفذ تمريرة أمامية على الإدخال tensor من خلال طبقات النموذج المبنية.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
def forward(self, x):
    """Executes a forward pass on the input tensor through the constructed model layers."""
    return self.forward_features(x)

forward_features(x)

يقوم بتشغيل المدخلات عبر طبقات النموذج وإرجاع المخرجات المحولة.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
725 726 727 728 729 730 731 732 733 734 735 736737738
def forward_features(self, x):
    """Runs the input through the model layers and returns the transformed output."""
    x = self.patch_embed(x)  # x input is (N, C, H, W)

    x = self.layers[0](x)
    start_i = 1

    for i in range(start_i, len(self.layers)):
        layer = self.layers[i]
        x = layer(x)
    B, _, C = x.shape
    x = x.view(B, 64, 64, C)
    x = x.permute(0, 3, 1, 2)
    return self.neck(x)

no_weight_decay_keywords()

إرجاع قاموس لأسماء المعلمات حيث لا يجب تطبيق اضمحلال الوزن.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
720 721 722 723
@torch.jit.ignore
def no_weight_decay_keywords(self):
    """Returns a dictionary of parameter names where weight decay should not be applied."""
    return {"attention_biases"}

set_layer_lr_decay(layer_lr_decay)

يضبط اضمحلال معدل التعلم لكل طبقة في نموذج TinyViT.

شفرة المصدر في ultralytics/models/sam/modules/tiny_encoder.py
674 675 676 677 678 679 680 681 682 683 684 685 686 687688689 690 691 692 693 694 695 696 697 698699 700 701702 703704705706707
def set_layer_lr_decay(self, layer_lr_decay):
    """Sets the learning rate decay for each layer in the TinyViT model."""
    decay_rate = layer_lr_decay

    # Layers -> blocks (depth)
    depth = sum(self.depths)
    lr_scales = [decay_rate ** (depth - i - 1) for i in range(depth)]

    def _set_lr_scale(m, scale):
        """Sets the learning rate scale for each layer in the model based on the layer's depth."""
        for p in m.parameters():
            p.lr_scale = scale

    self.patch_embed.apply(lambda x: _set_lr_scale(x, lr_scales[0]))
    i = 0
    for layer in self.layers:
        for block in layer.blocks:
            block.apply(lambda x: _set_lr_scale(x, lr_scales[i]))
            i += 1
        if layer.downsample is not None:
            layer.downsample.apply(lambda x: _set_lr_scale(x, lr_scales[i - 1]))
    assert i == depth
    for m in [self.norm_head, self.head]:
        m.apply(lambda x: _set_lr_scale(x, lr_scales[-1]))

    for k, p in self.named_parameters():
        p.param_name = k

    def _check_lr_scale(m):
        """Checks if the learning rate scale attribute is present in module's parameters."""
        for p in m.parameters():
            assert hasattr(p, "lr_scale"), p.param_name

    self.apply(_check_lr_scale)





تم إنشاء 2023-11-12, اخر تحديث 2023-11-25
المؤلفون: جلين جوشر (3) ، يضحك س (1)