cvnets.models.segmentation package

Subpackages

• cvnets.models.segmentation.heads package
  • Submodules
  • cvnets.models.segmentation.heads.base_seg_head module
    • BaseSegHead
      • BaseSegHead.__init__()
      • BaseSegHead.forward_aux_head()
      • BaseSegHead.forward_seg_head()
      • BaseSegHead.forward()
      • BaseSegHead.reset_head_parameters()
      • BaseSegHead.add_arguments()
      • BaseSegHead.get_trainable_parameters()
      • BaseSegHead.update_classifier()
      • BaseSegHead.build_model()
  • cvnets.models.segmentation.heads.deeplabv3 module
    • DeeplabV3
      • DeeplabV3.__init__()
      • DeeplabV3.update_classifier()
      • DeeplabV3.add_arguments()
      • DeeplabV3.forward_seg_head()
  • cvnets.models.segmentation.heads.pspnet module
    • PSPNet
      • PSPNet.__init__()
      • PSPNet.update_classifier()
      • PSPNet.add_arguments()
      • PSPNet.forward_seg_head()
  • cvnets.models.segmentation.heads.simple_seg_head module
    • SimpleSegHead
      • SimpleSegHead.__init__()
      • SimpleSegHead.add_arguments()
      • SimpleSegHead.forward_seg_head()
  • Module contents

Submodules

cvnets.models.segmentation.base_seg module

class cvnets.models.segmentation.base_seg.BaseSegmentation(opts, encoder: BaseImageEncoder, *args, **kwargs)

Bases: BaseAnyNNModel

Base class for segmentation networks.

Parameters:

• opts – Command-line arguments

• encoder – Image classification network

__init__(opts, encoder: BaseImageEncoder, *args, **kwargs) → None

Initializes internal Module state, shared by both nn.Module and ScriptModule.

classmethod add_arguments(parser: ArgumentParser) → ArgumentParser

Add segmentation model specific arguments

maybe_seg_norm_layer()

set_default_norm_layer()

dummy_input_and_label(batch_size: int) → Dict

Create dummy input and labels for CI/CD purposes. Child classes must override it if functionality is different.

update_classifier(opts, n_classes: int) → None

This function updates the classification layer in a model. Useful for finetuning purposes.

classmethod set_model_specific_opts_before_model_building(opts: Namespace, *args, **kwargs) → Dict[str, Any]

cvnets.models.segmentation.base_seg.set_model_specific_opts_before_model_building(opts: Namespace) → Dict[str, Any]

Override library-level defaults with model-specific default values.

Parameters:

opts – Command-line arguments

Returns:

A dictionary containing the name of arguments that are updated along with their original values. This dictionary is used in unset_model_specific_opts_after_model_building function to unset the model-specific to library-specific defaults.

cvnets.models.segmentation.base_seg.unset_model_specific_opts_after_model_building(opts: Namespace, default_opts_info: Dict[str, Any], *ars, **kwargs) → None

Given command-line arguments and a mapping of opts that needs to be unset, this function unsets the library-level defaults that were over-ridden previously in set_model_specific_opts_before_model_building.

cvnets.models.segmentation.enc_dec module

class cvnets.models.segmentation.enc_dec.SegEncoderDecoder(opts, encoder: BaseImageEncoder, seg_head, *args, **kwargs)

Bases: BaseSegmentation

This class defines a encoder-decoder architecture for the task of semantic segmentation. Different segmentation heads (e.g., PSPNet and DeepLabv3) can be used

Parameters:

• opts – command-line arguments

• encoder (BaseImageEncoder) – Backbone network (e.g., MobileViT or ResNet)

__init__(opts, encoder: BaseImageEncoder, seg_head, *args, **kwargs) → None

Initializes internal Module state, shared by both nn.Module and ScriptModule.

get_trainable_parameters(weight_decay: float | None = 0.0, no_decay_bn_filter_bias: bool | None = False, *args, **kwargs)

This function separates the parameters for backbone and segmentation head, so that different learning rates can be used for backbone and segmentation head

forward(x: Tensor, *args, **kwargs) → Tuple[Tensor, Tensor] | Tensor | Dict

Implement the model-specific forward function in sub-classes.

update_classifier(opts, n_classes: int) → None

This function updates the classification layer in a model. Useful for finetuning purposes.

classmethod build_model(opts: Namespace, *args, **kwargs) → BaseAnyNNModel

Build a model from command-line arguments. Sub-classes must implement this method

Parameters:

opts – Command-line arguments

…note::

This function is typically implemented in the base class for each task and implementation is reused by all models in that task.

Module contents