Get the buffer of a specific item. Args: item (str): The demanded item. ids (list[int]): The demanded ids.

Run Kalman filter correction step. Args: mean (ndarray): The predicted state's mean vector (8 dimensional). covarianc

copy the refs.

Convert a version string into a tuple of integers. This method is usually used for comparing two versions. For pre-release versions: alpha <

Extract the features of the input image. Args: img (Tensor): image of shape (N, C, H, W). Returns: tuple(Ten

Initialize the weights of modules in single object tracker.

Create a superset of inputs needed to run test or train batches. Args: input_shape (tuple): input batch dimensions n

Build model.

hdfs capable to check whether a file_path is exists

hdfs mkdir

Parse the raw outputs (losses) of the network. Args: losses (dict): Raw output of the network, which usually contain

Convert tracking/detection results to a list of numpy arrays. Args: bboxes (torch.Tensor | np.ndarray): shape (n, 5) labels (torc

hdfs copy

Print information about the VQA annotation file. :return:

Compute loss. Args: track_results (dict): it may contains the following keys: - 'pred_bboxes': bboxes of (N, num_

Calculate feature similarity from embeddings. Args: key_embeds (Tensor): Shape (N1, C). ref_embeds (Tensor): Shape (N2, C).

Load anns with the specified ids. :param ids (int array) : integer ids specifying anns :return: anns (object array) : loaded an

Normalization.

Load a configuration as a python module.

Calculate the ground truth for all samples in a batch according to the sampling_results. Args: sampling_results (List[obj

Restore the results (list of results of each category) into the results of the model forward. Args: bbox_results (list[np.ndarray]):

Parse person id annotations.

Use flow to warp feature map. Args: x (Tensor): of shape (N, C, H_x, W_x). flow (Tensor): of shape (N, C, H_f, W_f). Returns

compatible to HDFS path or local path

vision_embeds: cls + patch embeds

Reset the buffer of the tracker.

Save ckpt to local or HDFS

Build tracker.

Update the tracker. Args: kwargs (dict[str: Tensor | int]): The `str` indicates the name of the input variable. `

Convert bbox coordinates from (x1, y1, x2, y2) to (cx, cy, ratio, h). Args: bbox (Tensor): Shape (n, 4) for bboxes. Returns:

Build PyTorch DataLoader. In distributed training, each GPU/process has a dataloader. In non-distributed training, there is only one dataload

Get COCO annotations by the information of image. Args: img_info (int): Information of image. Returns: dict:

Get root logger. Args: log_file (str): File path of log. Defaults to None. log_level (int): The level of logger. Defaults to logg

Initialize a model from config file. Args: config (str or :obj:`mmcv.Config`): Config file path or the config object.

Prepare results dict for pipeline. The following keys in dict will be called in the subsequent pipeline.

Prepare sampled training data according to the sampled index. Args: video_info (dict): the video information. It contains the key

Draw `result` over `img`. Args: img (str or Tensor): The image to be displayed. result (dict): The results to draw ov

Test without augmentations.

Track the box `bbox` of previous frame to current frame `img`. Args: img (Tensor): of shape (1, C, H, W). bbox (list

Build motion model.

Crop the images according to some bounding boxes. Typically for re- identification sub-module. Args: img (Tensor): of sha

Format the results to json (standard format for TAO evaluation). Args: results (list[ndarray]): Testing results of the dataset.

open a file on hdfs with contextmanager. Args: mode (str): supports ["r", "w", "wa"]

Parse attribute of each video in VOT. Args: video_path (str): The path of video. attr_name (str): The name of video's attribute.

Parse the package dependencies listed in a requirements file but strips specific versioning information. Args: fname (str): path to r

Prepare sampled training data according to the sampled index. Args: video_info (dict): the video information. It contains the key

Step forward with the velocity.

Convert annotation which can be polygons, uncompressed RLE to RLE. :return: binary mask (numpy 2D array)

Extract roi features.

Get ann ids that satisfy given filter conditions. default skips that filter. :param vidIds (int array) : get anns for given vids

Get bboxes annotation about the instance in a video. Args: video_ind (int): video index Returns: ndarray: in