Return the learning rate warmup factor at a specific iteration. See https://arxiv.org/abs/1706.02677 for more details. Args: met

Prepare ._gts and ._dts for evaluation based on params :return: None

Based on the matching between N proposals and M groundtruth, sample the proposals and set their classification labels. Args:

Register a class name to that class ID.

Add a ground truth. If box or mask is None, this GT will be ignored for that mode.

Call `add_ground_truth_to_proposals_single_image` for all images. Args: gt_boxes(list[Boxes]): list of N elements. Element i is a Bo

Call this once per image.

aligned interpolation in feature-level. This interpolation method, writen by Adelaidet Condinst Codebase and beneficial to seg task, gain

Apply the transform on a list of polygons, each represented by a Nx2 array. By default will just transform all the points.

Computes area of the mask. Only works with Polygons, using the shoelace formula: https://stackoverflow.com/questions/24467972

Benchmark the input function and print out the results. Args: func (callable): a closure that returns a function for benchmarking,

dataset_names: List[str], in which elemements must be in format of "dataset_task_version"

Args: cfg: a cvpods CfgNode Returns: backbone (Backbone): backbone module, must be a subclass of :class:`Backbone`.

Check that the datasets have consistent metadata. Args: key (str): a metadata key dataset_names (list[str]): a list of datas

Performs all_gather operation on the provided tensors. *** Warning ***: torch.distributed.all_gather has no gradient.

Args: binary_mask (ndarray): numpy array of shape (H, W), where H is the image height and W is the image width. E

Args: text (str): class label position (tuple): a tuple of the x and y coordinates to place text on image.

Dynamic import a project. Args: config_name (str): module name config_path (str): the dir that contains the .py with this mo

A file lock. Once entered, it is guaranteed that no one else holds the same lock. Others trying to enter the lock will block for 30 minutes a

filter header and data in filter list. Args: filter_list (Iterable[string]): list of headers that needs to be filtered o

Returns a ClassedAPDataObject where all errors given the condition returns True are fixed.

Warning: result not cached.

Find all BatchNorm (BN) modules that are in training mode. See cvpods.modeling.nn_utils.precise_bn.BN_MODULE_TYPES for a list of all modules

insert header and data into current table at given position Args: header (Iterable[string]): inserted data.

A wrapper around :func:`torch.nn.functional.interpolate` to support zero-size tensor.

create a softlink to output dir. Args: link_name(str): name of softlink

SSD Weighted Loss Function: L(x,c,l,g) = (Lconf(x, c) + αLloc(x,l,g)) / N Where, Lconf is the CrossEntropy Loss and L

Return the losses from a set of RPN predictions and their associated ground-truth. Returns: dict[loss name -> loss value

Create a snet stage by creating many blocks. Args: num_blocks (int): the number of blocks in this stage. input_channels (int

Convert pred_mask_logits to estimated foreground probability masks while also extracting only the masks for the predicted classes in pred_ins

Compute the mask prediction loss defined in the Mask R-CNN paper. Args: pred_mask_logits (Tensor): A tensor of shape (B, C, Hmask, W

Utility function to parse official darknet weights into torch.

Rearrange the tensor layout from the network output, i.e.: list[Tensor]: #lvl tensors of shape (N, A x K, Hi, Wi) to per-image prediction

Get features from feature maps in `features_list` that correspond to specific point coordinates inside each bounding box from `boxes`.

Return objectness logits in the same format as the proposals returned by :meth:`predict_proposals`. Returns: pre

Returns: list[Tensor]: A list of Tensors of predicted class probabilities for each image. Element i has shape (Ri

Transform anchors into proposals by applying the predicted anchor deltas. Returns: proposals (list[Tensor]): A list of L

Print main metrics in a format similar to Detectron, so that they are easy to copypaste into a spreadsheet. Args: results (Order

Converting the points set into bounding box. :param pts: the input points sets (fields), each points set (fields) is rep

Args: box_scores (N, 5): boxes in corner-form and probabilities. (Note here 5 == 4 + 1, i.e., 4-dim horizontal bo

Register hooks to the runner. The hooks are executed in the order they are registered. Args: hooks (list[Optiona

Register the given function as a handler that this transform will use for a specific data type. Args: data_type

Preparation for a new round of evaluation. Should be called before starting a round of evaluation.

Return semantic segmentation predictions in the original resolution. The input images are often resized when entering semantic segmentor. Mo

Load an LVIS-style dataset. The version string is used for downloading the dataset and should be one of the versions of LVIS (e.g., v0.5,

Args: cfg (CfgNode): model (GeneralizedRCNN): a GeneralizedRCNN to apply TTA on. tta_mapper (callable): t

Args: bottom_up (Backbone): module representing the bottom up subnetwork. Must be a subclass of :class:`Backbone`

input_size (int): the input image size. bottom_up (Backbone): module representing the bottom up subnetwork. Must be a sub

Args: input_channels (int): the input channel number. output_channels (int): the output channel number. n

Args: input_size (int): channels, or (channels, height, width) num_classes (int): number of foreground classes

Poly LR with warmup Args: optimizer (torch.optim.Optimizer): optimizer used. max_iters (int): max num of iter

Assert that each stride is 2x times its preceding stride, i.e. "contiguous in log2".

Naive tracking heuristics to assign same color to the same instance, will update the internal state of tracked instances. Re

Execute inference on a list of inputs, using batch size = self.batch_size, instead of the length of the list. Inputs & outpu

Calculate proper im2col step size, which should be divisible by input_size and not larger than prefer_size. Meanwhile the step size s

In-place convert all numpy arrays in the state_dict to torch tensor. Args: state_dict (dict): a state-dict to be loaded t

Derive the desired score numbers from summarized COCOeval. Args: coco_eval (None or COCOEval): None represents no predic

Evaluate the box proposals in self._predictions. Fill self._results with the metrics for "box_proposals" task.

Evaluate self._predictions on the given tasks. Fill self._results with the metrics of the tasks.

Forward logic of the mask prediction branch. Args: features (dict[str, Tensor]): #level input features for mask predicti

Forward logic of the coarse mask head.

Forward logic of the mask point head.

Return an activation function given a string # TODO: move to layers

This func may be rewirite someday

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