Utility function for network_to_half(). Retained for legacy purposes.

Build a flat label space by creating the `universal` taxonomy here. Instead of using entries from a mapping TSV, we just count the universal

Build a flat label space by adding each `universal` taxonomy entry from the TSV here. We make a mapping from universal_name->universal_id, and

Checks that `replacement`, which is intended to replace `original` is of the right type. The type is correct if it matches exactly or is one of a

Create a grayscale version of the original image. The colors in masked area, if given, will be kept.

Args: classes (list[int] or None): scores (list[float] or None): class_names (list[str] or None): Returns: l

Decodes a raw config value (e.g., from a yaml config files or command line argument) into a Python object.

Cache conversion maps so we will later be able to perform fast grayscale->grayscale mapping for training data transformation. Each map is impl

Explicitly for inference on our test datasets. We implement this remapping from mu classes to mt classes for the evaluation dataset as a line

Identical to ResNet `_make_layer()`, except `inplanes` is an explicit argument rather than class attribute, and batch norm im

Computes the norm over all dimensions except dim

Args: Same as in :meth:`__init__()`. Returns: fig (matplotlib.pyplot.figure): top level container for all th

Transform a training dataset to the universal taxonomy. For one training dataset, map each of its class ids to one universal ids. Args: -

Explicitly for inference on our test datasets. Store correspondences between universal_taxonomy and test_dataset_taxonomy. Args: - dataset:

LR schedule that should yield 76% converged accuracy with batch size 256

Applies a given weight reparameterization (such as weight normalization) to a parameter in the given module. If no parameter is given, applie

should be perfect agreement between label image and instance image, then we will enforce strict boundary agreement between the two.

assign one value to all elements

* Builds the index */

Args: - relabeled_data_list - save_vis: whether to save visualize examplars

3x3 convolution with padding

Converts a module's immediate parameters and buffers to dtype.

# Convert the given annotation to a label image Args: - json_fpath - annotation Returns: