Skip to content

Computer Vision

tensoris.backend.components.layers.cv

Computer Vision Atomic Neural Network Layers.

Classes

ConvStem

Bases: Module

Convolutional Stem Layer for Computer Vision Backbones.

Source code in src/tensoris/backend/components/layers/cv.py 
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
class ConvStem(nn.Module):
    """Convolutional Stem Layer for Computer Vision Backbones."""

    def __init__(self, in_channels: int = 3, out_channels: int = 64) -> None:
        """Initialize ConvStem layer.

        Args:
            in_channels: Input image channels (3 for RGB).
            out_channels: Output feature map channels.
        """
        super().__init__()
        self.stem = nn.Sequential(
            nn.Conv2d(
                in_channels,
                out_channels,
                kernel_size=7,
                stride=2,
                padding=3,
                bias=False,
            ),
            nn.BatchNorm2d(out_channels),
            nn.ReLU(inplace=True),
            nn.MaxPool2d(kernel_size=3, stride=2, padding=1),
        )

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        """Execute ConvStem forward pass."""
        return self.stem(x)
Methods:
__init__ 
__init__(in_channels=3, out_channels=64)

Initialize ConvStem layer.

Parameters:

Name Type Description Default
in_channels int

Input image channels (3 for RGB).

 3
out_channels int

Output feature map channels.

 64
Source code in src/tensoris/backend/components/layers/cv.py 
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
def __init__(self, in_channels: int = 3, out_channels: int = 64) -> None:
    """Initialize ConvStem layer.

    Args:
        in_channels: Input image channels (3 for RGB).
        out_channels: Output feature map channels.
    """
    super().__init__()
    self.stem = nn.Sequential(
        nn.Conv2d(
            in_channels,
            out_channels,
            kernel_size=7,
            stride=2,
            padding=3,
            bias=False,
        ),
        nn.BatchNorm2d(out_channels),
        nn.ReLU(inplace=True),
        nn.MaxPool2d(kernel_size=3, stride=2, padding=1),
    )
forward 
forward(x)

Execute ConvStem forward pass.

Source code in src/tensoris/backend/components/layers/cv.py 
32
33
34
def forward(self, x: torch.Tensor) -> torch.Tensor:
    """Execute ConvStem forward pass."""
    return self.stem(x)