Helper function to synchronize (barrier) among all processes when using distributed training

Return a process group based on gloo backend, containing all the ranks The result is cached.

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

Computes area of boxes. Args: boxes: Numpy array with shape [N, 4] holding N boxes Returns: a numpy array with shape [N*1] representing

Computes area of masks. Args: masks: Numpy array with shape [N, height, width] holding N masks. Masks values are of type np.uint8 and val

:param videos: a list of video tensors :param size_divisible: output_size(width and height) should be divisble by this param :return: bat

Blend two images with a given weight alpha. Args: images1 (tensor): the first images to be blended, the dimension is `num

encode text. Args: text (dict): The output of huggingface's `PreTrainedTokenizer`. contains keys: - input_ids (tor

Evaluates detections and returns a dictionary of metrics.

calculate pair-wise video-text similarity. Args: vision_proj (torch.Tensor): The vision representation. Shape: [B,T,C]. text_proj

Sinusoid position encoding table

Perform horizontal flip on the given images and corresponding boxes. Args: prob (float): probility to flip the images. images

Returns a unique identifier for a video id & timestamp.

Returns a unique identifier for a video id & timestamp.

Perform random spatial crop on the given images and corresponding boxes. Args: images (tensor): images to perform random crop. The di

Basic Tokenization of a piece of text. Split on "white spaces" only, for sub-word tokenization, see WordPieceTokenizer. Args:

Perform uniform spatial sampling on the images and corresponding boxes. Args: images (tensor): images to perform uniform crop. The di

Runs basic whitespace cleaning and splitting on a piece of text.

Copies the extra fields of boxlist_to_copy_from to boxlist_to_copy_to. Args: boxlist_to_copy_to: BoxList to which extra fields are copied.

Returns numpy arrays belonging to class with index `class_index`. Args: detected_boxes: A numpy array containing detected boxes. dete

Workaround for ModelEma._load_checkpoint to accept an already-loaded object

Workaround for ModelEma._load_checkpoint to accept an already-loaded object

Returns: list[int]: size of the tensor, on each rank Tensor: padded tensor that has the max size

Strips accents from a piece of text.

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

Converts (H, W, C) numpy.ndarray to (C, W, H) format

Peform crop on the bounding boxes given the offsets. Args: boxes (ndarray or None): bounding boxes to peform crop. The dimension

Filter to keep only boxes and masks with score exceeding a given threshold. This op keeps the collection of boxes and masks whose corresponding sco

Filter to keep only boxes with score exceeding a given threshold. This op keeps the collection of boxes whose corresponding scores are greater th

flatten a list of lists [[1,2], [3,4]] to [1,2,3,4]

Args: sim (torch.Tensor): The similarity between videos and texts. shape: (B, B). idx (torch.Tensor): The index for e

get the media types for for all the dataloaders. Args: datasources (List): List of dataloaders or datasets. Returns: List. The media