Stores necessary information to remap a list of (segment, image) objects from an old class to a new class.

Computes and stores the average and current value

Computes and stores the average and current value

Multi-headed attention. See "Attention Is All You Need" for more details.

Machine-generated taxonomy, generated without domain knowledge. Inherits from the `TaxonomyConverter` class, and re-uses most functionality. Input

Multi-headed attention. See "Attention Is All You Need" for more details.

We use 1x1 convolution for our linear mapping from universal->test_taxonomy. Private methods marked with leading underscore.

Simple API to interact with instance masks of ADE20K dataset.

Simple API to access masks of Cityscapes and IDD. Format is shared almost exactly between Cityscapes and IDD, except IDD sequence IDs are integer

r"""Applies Layer Normalization over a mini-batch of inputs as described in the paper `Layer Normalization`_ . Currently only runs on cuda()

r"""Implements stochastic gradient descent (optionally with momentum). Currently GPU-only. Requires Apex to be installed via ``pip install -

Launch MLP in C++ Args: mlp_sizes (list of int): MLP sizes. Example: [1024,1024,1024] will create 2 MLP layers with shape 1024x1024

Computes and stores the average and current value

CfgNode represents an internal node in the configuration tree. It's a simple dict-like container that allows for attribute-based access to ke

Implements Adam algorithm. Currently GPU-only. Requires Apex to be installed via ``python setup.py install --cuda_ext --cpp_ext``. It has be

RNNCell gate_multiplier is related to the architecture you're working with For LSTM-like it will be 4 and GRU-like will be 3. Alway

Simple API to interact with instance masks of COCO Panoptic dataset. We take the instance ID images from the Panoptic .png files, and we take the

Class that manages a static loss scale. This class is intended to interact with :class:`FP16_Optimizer`, and should not be directly manipula

SuperPoint Convolutional Detector and Descriptor SuperPoint: Self-Supervised Interest Point Detection and Description. Daniel DeTone, Tomasz

stackedRNN

Class to handle convert operations.

Class to store all the data for every kernel e.g. name, bytes, flops, device, stream etc.

:class:`apex.parallel.DistributedDataParallel` is a module wrapper that enables easy multiprocess distributed data parallel training, similar

Implements Adam algorithm. Currently GPU-only. Requires Apex to be installed via ``pip install -v --no-cache-dir --global-option="--cpp_ext"

Attribute: polygons (list[ndarray]): list[ndarray]: polygons for this mask. Each ndarray has format [x, y, x, y, ...]

Tree node

synchronized batch normalization module extented from ``torch.nn.BatchNormNd`` with the added stats reduction across multiple processes.

This class handles the various activation functions.

# N = batch size # C,H,W = input channels, height, width # K,P,Q = output channels, height, width # R,S = filter height, width # g = groups

This class provides functions for DB operations with exception handling.

Convert model to half precision in a batchnorm-safe way.

:class:`FP16_Optimizer` A cutdown version of apex.fp16_utils.FP16_Optimizer. Designed only to wrap apex.contrib.optimizers.FusedAdam, FusedSG

An object of Foo is instantiated when we detect an unsupported operator.

PoseNet using Inception V3