cvnets.models package

Subpackages

• cvnets.models.audio_classification package
  • Submodules
  • cvnets.models.audio_classification.audio_byteformer module
    • AudioByteFormer
      • AudioByteFormer.forward()
      • AudioByteFormer.dummy_input_and_label()
  • cvnets.models.audio_classification.base_audio_classification module
    • BaseAudioClassification
      • BaseAudioClassification.__init__()
      • BaseAudioClassification.add_arguments()
  • Module contents
• cvnets.models.classification package
  • Subpackages
    • cvnets.models.classification.config package
      • Submodules
      • cvnets.models.classification.config.byteformer module
      • cvnets.models.classification.config.efficientnet module
      • cvnets.models.classification.config.fastvit module
      • cvnets.models.classification.config.mobilenetv1 module
      • cvnets.models.classification.config.mobilenetv2 module
      • cvnets.models.classification.config.mobilenetv3 module
      • cvnets.models.classification.config.mobileone module
      • cvnets.models.classification.config.mobilevit module
      • cvnets.models.classification.config.mobilevit_v2 module
      • cvnets.models.classification.config.regnet module
      • cvnets.models.classification.config.resnet module
      • cvnets.models.classification.config.swin_transformer module
      • cvnets.models.classification.config.vit module
      • Module contents
  • Submodules
  • cvnets.models.classification.base_image_encoder module
    • BaseImageEncoder
      • BaseImageEncoder.__init__()
      • BaseImageEncoder.add_arguments()
      • BaseImageEncoder.check_model()
      • BaseImageEncoder.update_classifier()
      • BaseImageEncoder.extract_end_points_all()
      • BaseImageEncoder.extract_end_points_l4()
      • BaseImageEncoder.extract_features()
      • BaseImageEncoder.forward_classifier()
      • BaseImageEncoder.forward()
      • BaseImageEncoder.get_trainable_parameters()
      • BaseImageEncoder.dummy_input_and_label()
      • BaseImageEncoder.get_exportable_model()
      • BaseImageEncoder.build_model()
    • set_model_specific_opts_before_model_building()
    • unset_model_specific_opts_after_model_building()
  • cvnets.models.classification.byteformer module
    • unfold_tokens()
    • ByteFormer
      • ByteFormer.__init__()
      • ByteFormer.add_arguments()
      • ByteFormer.dummy_input_and_label()
      • ByteFormer.apply_token_reduction_net()
      • ByteFormer.get_backbone_inputs()
      • ByteFormer.backbone_forward()
      • ByteFormer.get_downsampler_name()
      • ByteFormer.get_downsampler()
      • ByteFormer.forward()
      • ByteFormer.build_model()
  • cvnets.models.classification.efficientnet module
    • EfficientNet
      • EfficientNet.__init__()
      • EfficientNet.add_arguments()
  • cvnets.models.classification.fastvit module
    • basic_blocks()
    • FastViT
      • FastViT.__init__()
      • FastViT.add_arguments()
      • FastViT.get_exportable_model()
  • cvnets.models.classification.mobilenetv1 module
    • MobileNetv1
      • MobileNetv1.__init__()
      • MobileNetv1.add_arguments()
  • cvnets.models.classification.mobilenetv2 module
    • MobileNetV2
      • MobileNetV2.__init__()
      • MobileNetV2.add_arguments()
  • cvnets.models.classification.mobilenetv3 module
    • MobileNetV3
      • MobileNetV3.__init__()
      • MobileNetV3.add_arguments()
  • cvnets.models.classification.mobileone module
    • MobileOne
      • MobileOne.__init__()
      • MobileOne.add_arguments()
      • MobileOne.get_exportable_model()
  • cvnets.models.classification.mobilevit module
    • MobileViT
      • MobileViT.__init__()
      • MobileViT.add_arguments()
  • cvnets.models.classification.mobilevit_v2 module
    • MobileViTv2
      • MobileViTv2.__init__()
      • MobileViTv2.add_arguments()
  • cvnets.models.classification.regnet module
    • RegNet
      • RegNet.__init__()
      • RegNet.add_arguments()
  • cvnets.models.classification.resnet module
    • ResNet
      • ResNet.__init__()
      • ResNet.add_arguments()
  • cvnets.models.classification.swin_transformer module
    • SwinTransformer
      • SwinTransformer.__init__()
      • SwinTransformer.extract_end_points_all()
      • SwinTransformer.add_arguments()
  • cvnets.models.classification.vit module
    • VisionTransformer
      • VisionTransformer.__init__()
      • VisionTransformer.update_layer_norm_eps()
      • VisionTransformer.reset_simple_fpn_params()
      • VisionTransformer.add_arguments()
      • VisionTransformer.extract_patch_embeddings()
      • VisionTransformer.extract_features()
      • VisionTransformer.forward_classifier()
      • VisionTransformer.forward()
      • VisionTransformer.extract_end_points_all()
  • Module contents
• cvnets.models.detection package
  • Subpackages
    • cvnets.models.detection.utils package
      • Submodules
      • cvnets.models.detection.utils.rcnn_utils module
      • Module contents
  • Submodules
  • cvnets.models.detection.base_detection module
    • BaseDetection
      • BaseDetection.__init__()
      • BaseDetection.add_arguments()
      • BaseDetection.reset_layer_parameters()
      • BaseDetection.build_model()
    • check_feature_map_output_channels()
  • cvnets.models.detection.mask_rcnn module
    • MaskRCNNEncoder
      • MaskRCNNEncoder.__init__()
      • MaskRCNNEncoder.get_augmented_tensor()
      • MaskRCNNEncoder.forward()
      • MaskRCNNEncoder.get_trainable_parameters()
    • MaskRCNNDetector
      • MaskRCNNDetector.__init__()
      • MaskRCNNDetector.update_layer_norm_eps()
      • MaskRCNNDetector.set_norm_layer_opts()
      • MaskRCNNDetector.reset_norm_layer_opts()
      • MaskRCNNDetector.add_arguments()
      • MaskRCNNDetector.reset_generalized_rcnn_transform()
      • MaskRCNNDetector.get_trainable_parameters()
      • MaskRCNNDetector.forward()
      • MaskRCNNDetector.predict()
      • MaskRCNNDetector.dummy_input_and_label()
  • cvnets.models.detection.ssd module
    • SingleShotMaskDetector
      • SingleShotMaskDetector.coordinates
      • SingleShotMaskDetector.__init__()
      • SingleShotMaskDetector.add_arguments()
      • SingleShotMaskDetector.reset_layers()
      • SingleShotMaskDetector.process_anchors_ar()
      • SingleShotMaskDetector.get_backbone_features()
      • SingleShotMaskDetector.ssd_forward()
      • SingleShotMaskDetector.forward()
      • SingleShotMaskDetector.predict()
      • SingleShotMaskDetector.postprocess_detections()
      • SingleShotMaskDetector.dummy_input_and_label()
  • Module contents
    • DetectionPredTuple
      • DetectionPredTuple.boxes
      • DetectionPredTuple.labels
      • DetectionPredTuple.masks
      • DetectionPredTuple.scores
• cvnets.models.multi_modal_img_text package
  • Submodules
  • cvnets.models.multi_modal_img_text.base_multi_modal_img_text module
    • BaseMultiModalImageText
      • BaseMultiModalImageText.__init__()
      • BaseMultiModalImageText.add_arguments()
  • cvnets.models.multi_modal_img_text.clip module
    • CLIP
      • CLIP.__init__()
      • CLIP.add_arguments()
      • CLIP.reset_parameters()
      • CLIP.get_trainable_parameters()
      • CLIP.dummy_input_and_label()
      • CLIP.forward()
      • CLIP.build_model()
    • update_image_classifier()
  • Module contents
• cvnets.models.segmentation package
  • Subpackages
    • cvnets.models.segmentation.heads package
      • Submodules
      • cvnets.models.segmentation.heads.base_seg_head module
      • cvnets.models.segmentation.heads.deeplabv3 module
      • cvnets.models.segmentation.heads.pspnet module
      • cvnets.models.segmentation.heads.simple_seg_head module
      • Module contents
  • Submodules
  • cvnets.models.segmentation.base_seg module
    • BaseSegmentation
      • BaseSegmentation.__init__()
      • BaseSegmentation.add_arguments()
      • BaseSegmentation.maybe_seg_norm_layer()
      • BaseSegmentation.set_default_norm_layer()
      • BaseSegmentation.dummy_input_and_label()
      • BaseSegmentation.update_classifier()
      • BaseSegmentation.set_model_specific_opts_before_model_building()
    • set_model_specific_opts_before_model_building()
    • unset_model_specific_opts_after_model_building()
  • cvnets.models.segmentation.enc_dec module
    • SegEncoderDecoder
      • SegEncoderDecoder.__init__()
      • SegEncoderDecoder.get_trainable_parameters()
      • SegEncoderDecoder.forward()
      • SegEncoderDecoder.update_classifier()
      • SegEncoderDecoder.build_model()
  • Module contents

Submodules

cvnets.models.base_model module

class cvnets.models.base_model.BaseAnyNNModel(opts, *args, **kwargs)

Bases: Module

Base class for any neural network

__init__(opts, *args, **kwargs) → None

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

classmethod add_arguments(parser: ArgumentParser)

Add model-specific arguments

reset_parameters(opts: Namespace) → None

Initialize model weights

forward(x: Any, *args, **kwargs) → Any

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

get_trainable_parameters(weight_decay: float = 0.0, no_decay_bn_filter_bias: bool = False, module_name: str = '', *args, **kwargs) → Tuple[List[Mapping], List[float]]

Get parameters for training along with the learning rate.

Parameters:

• weight_decay – weight decay

• no_decay_bn_filter_bias – Do not decay BN and biases. Defaults to False.

Returns:

Returns a tuple of length 2. The first entry is a list of dictionary with three keys (params, weight_decay, param_names). The second entry is a list of floats containing learning rate for each parameter.

Note

Kwargs may contain module_name. To avoid multiple arguments with the same name, we pop it and concatenate with encoder or head name

dummy_input_and_label(batch_size: int) → Dict

Create dummy input and labels for CI/CD purposes. Child classes should implement it.

get_exportable_model() → Module

This function can be used to prepare the architecture for inference. For example, re-parameterizing branches when possible. The functionality of this method may vary from model to model, so child model classes have to implement this method, if such a transformation exists.

classmethod freeze_norm_layers(opts: Namespace, model: BaseAnyNNModel) → None

Freeze normalization layers in the model

Parameters:

• opts – Command-line arguments

• model – An instance of BaseAnyNNModel

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.

info() → None

Prints model, parameters, and FLOPs on start rank.

Module contents

cvnets.models.get_model(opts: Namespace, category: str | None = None, model_name: str | None = None, *args, **kwargs) → BaseAnyNNModel

Create a task-specific model from command-line arguments. If model category (or task) and name are passed as an argument, then they are used. Otherwise, dataset.category and model.{category}.name are read from command-line arguments to read model category and name, respectively.

Parameters:

• opts – Command-line arguments

• category – Category or task (e.g., segmentation)

• model_name – Model name for a specific task (e.g., vit for classification)

Returns:

An instance of cvnets.models.BaseAnyNNModel.

cvnets.models.arguments_model(parser: ArgumentParser) → ArgumentParser