sadimanna

reg_loss = 0
for mod in self.model.modules():
    if isinstance(mod, _BatchNorm):
        if self.decay_bn:
            for name, param in mod.named_parameters(recurse=False):
                reg_loss = reg_loss + param.norm(2)    
    else:
        for name, param in mod.named_parameters(recurse=False):
            if not name.endswith("bias"):
                reg_loss = reg_loss + param.norm(2)

sadimanna

@tf.keras.saving.register_keras_serializable(name="weighted_categorical_crossentropy")
def weighted_categorical_crossentropy(target, output, weights, axis=-1):
    target = tf.convert_to_tensor(target)
    output = tf.convert_to_tensor(output)
    target.shape.assert_is_compatible_with(output.shape)
    weights = tf.reshape(tf.convert_to_tensor(weights, dtype=target.dtype), (1,-1))

    # Adjust the predictions so that the probability of
    # each class for every sample adds up to 1
    # This is needed to ensure that the cross entropy is

sadimanna

@tf.keras.saving.register_keras_serializable(name="weighted_binary_crossentropy")
def weighted_binary_crossentropy(target, output, weights):
    target = tf.convert_to_tensor(target)
    output = tf.convert_to_tensor(output)
    weights = tf.convert_to_tensor(weights, dtype=target.dtype)

    epsilon_ = tf.constant(tf.keras.backend.epsilon(), output.dtype.base_dtype)
    output = tf.clip_by_value(output, epsilon_, 1.0 - epsilon_)

    # Compute cross entropy from probabilities.

sadimanna

@tf.keras.saving.register_keras_serializable(name="WeightedCategoricalCrossentropy")
class WeightedCategoricalCrossentropy:
    def __init__(
        self,
        weights,
        label_smoothing=0.0,
        axis=-1,
        name="weighted_categorical_crossentropy",
        fn = None,
    ):

sadimanna

@tf.keras.saving.register_keras_serializable(name="WeightedBinaryCrossentropy")
class WeightedBinaryCrossentropy:
    def __init__(
        self,
        label_smoothing=0.0,
        weights = [1.0, 1.0],
        axis=-1,
        name="weighted_binary_crossentropy",
        fn = None,
    ):

sadimanna

@echo off
setlocal enabledelayedexpansion

REM Set the paths for the main image folder, train folder, val folder, and test folder
set "image_folder=C:\Path\to\ImageFolder"
set "train_folder=C:\Path\to\train"
set "val_folder=C:\Path\to\val"
set "test_folder=C:\Path\to\test"

REM Create the train, val, and test folders if they don't exist

sadimanna

REM Set the paths for Folder A, train folder, val folder, and test folder 
set "mask_folder_path=%base_path%\annotations\trimaps" 
set "train_mask_path=%base_path%\train_masks" 
REM Create the target folder if it doesn't exist 
mkdir "%train_mask_path%" 
set "extension=.EXT" 
REM Move label files from the label folder to the train folder based on file names 
for %%F in ("%train_image_path%\*") do ( 
    set "file_name=%%~nF" 
    move "%mask_folder_path%\!file_name!%extension%" "%train_mask_path%\" 

sadimanna

@echo off 
setlocal enabledelayedexpansion 
 
REM Set the paths for Folder A, train folder, val folder, and test folder 
set "base_path=C:\Users\ISI_UTS\Siladittya\MIDA2023" 
set "image_folder_path=%base_path%\images" 
set "train_image_path=%base_path%\train_images" 
set "val_image_path=%base_path%\val_images" 
set "test_image_path=%base_path%\test_images" 
 

sadimanna

#!/bin/bash

# Set the paths for the train folder, label folder, and target folder
train_folder_path="/path/to/train"
label_folder_path="/path/to/label"
target_folder_path="/path/to/target"
label_ext = ".labext"

# Create the target folder if it doesn't exist
mkdir -p "$target_folder_path"

sadimanna

#!/bin/bash

# Set the paths for Folder A and the three target folders
folder_a_path="/path/to/FolderA"
train_folder_path="/path/to/train"
val_folder_path="/path/to/val"
test_folder_path="/path/to/test"

# Create the target folders if they don't exist
mkdir -p "$train_folder_path"