Constructs box collection. Args: data: a numpy array of shape [N, 4] representing box coordinates Raises: ValueError: if bbox da

Initialized PerImageEvaluation by evaluation parameters. Args: num_groundtruth_classes: Number of ground truth object classes matchin

Constructs box collection. Args: box_data: a numpy array of shape [N, 4] representing box coordinates mask_data: a numpy array of sha

Constructor. Args: categories: A list of dicts, each of which has the following keys - 'id': (required) an integer id uniquely iden

Constructor. Args: categories: A list of dicts, each of which has the following keys - 'id': (required) an integer id uniquely iden

Args: dim (int): number of dimension for the input. dim_inner (int): number of dimension inside of the Non-local bloc

Naive version of Synchronized 3D BatchNorm. Args: num_sync_devices (int): number of device to sync. args (lis

Args: dim_in (int): the channel dimension of the input. fusion_conv_channel_ratio (int): channel ratio for the convol

The `__init__` method of any subclass should also contain these arguments. Args: cfg (CfgNode): model buildin

Args: reduction (str): specifies reduction to apply to the output. It can be "mean" (default) or "none".

Args: dim_in (int): the channel dimensions of the input. dim_out (int): the channel dimension of the output.

Args: dim_in (int): the channel dimensions of the input. dim_out (int): the channel dimension of the output.

Args: dim_in (int): the channel dimensions of the input. dim_out (int): the channel dimension of the output.

The `__init__` method of any subclass should also contain these arguments. ResStage builds p streams, where p can be greater or equal

The `__init__` method of any subclass should also contain these arguments. List size of 1 for single pathway models (C2D, I3D, Slow

The `__init__` method of any subclass should also contain these arguments. Args: dim_in (int): the channel dimension of

The `__init__` method of any subclass should also contain these arguments. ResNetBasicHead takes p pathways as input wher

Perform linear projection and activation as head for tranformers. Args: dim_in (int): the channel dimension of the input

Returns: (int): the number of videos in the dataset.

Adds detections for a single image to be used for evaluation. Args: image_id: A unique string/integer identifier for the image. detec

Adds detections for a single image to be used for evaluation. Args: image_key: A unique string/integer identifier for the image. dete

Adds groundtruth for a single image to be used for evaluation. Args: image_id: A unique string/integer identifier for the image. grou

Adds groundtruth for a single image to be used for evaluation. Args: image_id: A unique string/integer identifier for the image. grou

Adds groundtruth for a single image to be used for evaluation. Args: image_key: A unique string/integer identifier for the image. gro

Add a new scalar value to the deque.

Recursively find all SubBN modules and aggregate sub-BN stats. Args: module (nn.Module) Returns: count (int): number of S

All gathers the provided tensors from all processes across machines. Args: tensors (list): tensors to perform all gather across all p

Run all_gather on arbitrary picklable data (not necessarily tensors). Args: data: any picklable object group: a torch proces

All reduce the provided tensors from all processes across machines. Args: tensors (list): tensors to perform all reduce across all pr

Computes area of masks. Args: box_mask_list: np_box_mask_list.BoxMaskList holding N boxes and masks Returns: a numpy array with shape [N

Construct binary label vector given a list of label indices. Args: labels (list): The input label list. num_classes (int): Nu

Perform backwarding, gathering the gradients across different process/ GPU group.

Benchmark the speed of data loading in PySlowFast. Args: cfg (CfgNode): configs. Details can be found in slowfast/config

Change coordinate frame of the boxlist to be relative to window's frame. Given a window of the form [ymin, xmin, ymax, xmax], changes bounding bo

Determine whether the loss is NaN (not a number). Args: loss (loss): loss to check whether is NaN.

Clears the state to prepare for a fresh evaluation.

Clip an array of boxes to an image with the given height and width. Args: boxes (ndarray): bounding boxes to perform clipping.

Clip the boxes with the height and width of the image size. Args: boxes (ndarray): bounding boxes to peform crop. The dimension is

Clip bounding boxes to a window. This op clips input bounding boxes (represented by bounding box corners) to a window, optionally filtering out b

Perform color jitter on the given image. Args: image (array): image to perform color jitter. img_brightness (float): jitter r

Perform color jitter on the list of images. Args: images (list): list of images to perform color jitter. img_brightness (floa

Perform color nomration on the given images. Args: images (tensor): images to perform color normalization. Dimension is `

Perform color normalization on the image with the given mean and stddev. Args: image (array): image to perform color normalization.

Compute and update the batch norm stats to make it more precise. During training both bn stats and the weight are changing after every iterat

Compute Average Precision according to the definition in VOCdevkit. Precision is modified to ensure that it does not decrease as recall decrease.

Compute CorLoc according to the definition in the following paper. https://www.robots.ox.ac.uk/~vgg/rg/papers/deselaers-eccv10.pdf Returns nans

Compute precision and recall. Args: scores: A float numpy array representing detection score labels: A boolean numpy array representing tru

Weighted loss function

Concatenate list of box_mask_lists. This op concatenates a list of input box_mask_lists into a larger box_mask_list. It also handles concatena

Constructs the data loader for the given dataset. Args: cfg (CfgNode): configs. Details can be found in slowfast/config/d

Construct a stochastic gradient descent or ADAM optimizer with momentum. Details can be found in: Herbert Robbins, and Sutton Monro. "A s

Convert the caffe2_layer_name to pytorch format by apply the list of regular expressions. Args: caffe2_layer_name

Aggregate annotations from all frames of a video to form video-level labels. Args: labels (list): The input label list. Returns:

Compute the system memory (RAM) usage for the current device (GB). Returns: usage (float): used memory (GB). total (float): t

Reads a label map and returns a category index. Args: label_map_path: Path to `StringIntLabelMap` proto text file. Returns: A category i

Creates a category index with a single `object` class.

Create sampler for the given dataset. Args: dataset (torch.utils.data.Dataset): the given dataset. shuffle (bool): set to ``T

Collate function for detection task. Concatanate bboxes, labels and metadata from different samples in the first dimension instead of sta

Compute evaluation result. Returns: A dictionary of metrics with the following fields - 1. summary_metrics: 'Precision/mAP@<

Compute evaluation result. Returns: A named tuple with the following fields - average_precision: float numpy array of average preci

Forward function. Args: x: input features with shape of (num_windows*B, N, C) mask: (0/-inf) mask with shape of (num_