Expand a hierarchical dict of scalars into a flat dict of scalars. If results[k1][k2][k3] = v, the returned dict will have the entry {"k1

Use the given boxes in `instances` to produce other (non-box) per-ROI outputs. This is useful for downstream tasks where a box is kn

Look for keys with the following patterns: 1) Starts with "fpn.inner." Example: "fpn.inner.res2.2.sum.lateral.weight"

Freeze the first several stages of the ResNet. Commonly used in fine-tuning. Layers that produce the same feature map spatia

Freeze the first several stages of the model. Commonly used in fine-tuning. Layers that produce the same feature map spatial size ar

A context manager that annotates the "training" attribute of every submodule to constant, so that the training codepath in these modules can

Remove AliasWithName placeholder and rename the input/output of it

In-place fuse extra copy ops between cpu/gpu for the following case: a -CopyAToB-> b -CopyBToA> c1 -NextOp1-> d1

Helper for building a global average pooling layer.

Generate a tensor storing canonical anchor boxes, which are all anchor boxes of different sizes and aspect_ratios centered at (0, 0).

Generate a tensor storing canonical anchor boxes, which are all anchor boxes of different sizes, aspect_ratios, angles centered at (0

Generates per stage widths and depths from RegNet parameters.

Returns: Boxes: tight bounding boxes around bitmasks. If a mask is empty, it's bounding box will be all zero.

Convert pytorch-style structured inputs to caffe2-style inputs that are tuples of tensors. Args: batched_inputs

Returns path to a builtin config file. Args: config_path (str): config file name relative to detectron2's "configs/" dir

Args: gt_classes: a long tensor of shape R that contains the gt class label of each proposal. num_fed_loss_classes: m

Compute the output size given input size and target short edge length.

Creates a function that converts outputs of the caffe2 model to detectron2's standard format. The function uses information i

Returns the color map for visualizing the segmentation mask. Args: num_cls: Number of classes Returns: The color map

Take the output blobs from init_net by running it. Outputs: params: dict from blob name to numpy array device_options: dict f

Execute the policy based on input data, and decide what transform to apply to inputs. Args: args: Any fixed-length posit

For ONNX exported model, GroupNorm will be represented as ATen op, this can be a drop in replacement from ATen to GroupNorm

Guess the version of a partial config where the VERSION field is not specified. Returns the version, or the latest if cannot make a guess.

Returns True, if at least one inferense-based loader must be instantiated for training

Extract predicted keypoint locations from heatmaps. Args: maps (Tensor): (#ROIs, #keypoints, POOL_H, POOL_W). The predicted heatmap

Idenfity the reshape sub-graph in a protobuf. The reshape sub-graph is defined as matching the following pattern: (input_blob) -> Op_1 -

Run inference on the given inputs. Args: batched_inputs (list[dict]): same as in :meth:`forward` do_postproc

A context where the model is temporarily changed to eval mode, and restored to previous mode afterwards. Args: model: a torch Mo

Identical to :meth:`RetinaNet.inference_single_image.

Single-image inference. Return bounding-box detection results by thresholding on scores and applying non-maximum suppression (NMS).

Args: box_size (height, width): Size of the reference box. boundary_threshold (int): Boxes that extend beyond the ref

Returns a bool indicating whether torch.fx is currently symbolically tracing a module. Can be useful for gating module logic that is incompat

Benchmark an iterator/iterable for `num_iter` iterations with an extra `warmup` iterations of warmup. End early if `max_time_seconds` tim

Post process each predicted keypoint heatmap in `pred_keypoint_logits` into (x, y, score) and add it to the `pred_instances` as a `pred_k

Arguments: pred_keypoint_logits (Tensor): A tensor of shape (N, K, S, S) where N is the total number of instances in the batc

Same interface as :meth:`RetinaNet.label_anchors`, but implemented with FCOS anchor matching rule. Unlike RetinaNet, there a

Args: anchors (list[Boxes]): A list of #feature level Boxes. The Boxes contains anchors of this image on the spec

Args: anchors (list[Boxes]): anchors for each feature map. gt_instances: the ground-truth instances for each image.

Neural network layers that makes predictions from regional input features.

Neural network layers that makes predictions from input features.

Traverses all keyframes of a video file. Returns a list of keyframe timestamps. Timestamps are counts in timebase units. Args: vi

Args: image_dir (str): path to the raw dataset. e.g., "~/cityscapes/leftImg8bit/train". gt_dir (str): path to the raw annotations

Loads a JSON file with annotations in COCO instances format. Replaces `detectron2.data.datasets.coco.load_coco_json` to handle metadata i

Loads a JSON file with annotations in LVIS instances format. Replaces `detectron2.data.datasets.coco.load_lvis_json` to handle metadata i

Load Pascal VOC detection annotations to Detectron2 format. Args: dirname: Contain "Annotations", "ImageSets", "JPEGImages"