Creates a Classifications instance from a [timm](https://huggingface.co/docs/hub/timm) inference result. Args: t

Creates a Detections instance from a [YOLOv5](https://github.com/ultralytics/yolov5) inference result. Args: yol

Searches for the first string in `candidates` whose edit distance to `query` is less than or equal to `threshold`. Args: candida

Get category ids that satisfy given filter conditions. Args: cat_names: names of the categories to retrieve.

Return the smoothed `Detections` for a single track. Averages `xyxy` over all valid (non-`None`) frames in the track window. `confide

Create track from unassociated measurement. Args: measurement: The initial measurement vector. Returns:

Return ``True`` if ``rle`` is a COCO compressed RLE (``str`` or ``bytes``). Use this to branch between the compressed-string pipeline (:func:

Returns `True` if the `KeyPoints` object is considered empty. Returns: `True` if the object is empty, `False` otherwise.

Compares the metadata payloads of two Detections instances. Args: metadata_a, metadata_b: The metadata payloads of the instances.

Checks if a given string is a valid hex color. Args: hex_color: A hex color string with an optional leading "#". Supports

Loads PASCAL VOC XML annotations and returns the image name, a Detections instance, and a list of class names. Args: images_

Load prediction result into an EvaluationDataset object. Args: predictions: prediction result. Returns:

Load polygon zone configurations from a JSON file. This function reads a JSON file which contains polygon coordinates, and converts them

Load polygon zone configurations from a JSON file. This function reads a JSON file which contains polygon coordinates, and converts them

Return ``(xyxy, masks_dense, class_ids)`` with random polygon masks. *num_vertices* controls mask complexity: more vertices → jaggier boundary.

Random BGR image.

Args: detections: Detections object to be moved. offset: An array of shape `(2,)` containing offset values in the for

Args: xyxyxyxy: An array of shape `(n, 4, 2)` containing the oriented bounding boxes coordinates in format `[[x1, y1], [x

Run Kalman filter prediction step (Vectorized version). Args: mean: The object state means at the previous time step.

Compute the area of N oriented bounding boxes using the shoelace formula. Args: corners: OBB corner coordinates with shape `(N, 4, 2)`.

Open the JSON file for writing.

Convert detections and optional custom data into per-detection rows. Builds one dictionary per detection containing bounding box coo

Convert detections and optional custom data into per-detection rows. Builds one dictionary per detection containing bounding box coo

Convert a PNG byte string to a label mask array. Args: png_string: A byte string representing the PNG image. Returns: A

Process the result of Transformers object detection functions such as `post_process` (v4) and `post_process_detection` (v5). Args:

Process the result of the Transformers function `post_process_panoptic` (v4). Args: segmentation_result: Dictionary containing segme

Process the result of Transformers segmentation functions such as `post_process_panoptic`, `post_process_segmentation`, and `post_process_ins

Process the result of the Transformers function `post_process_panoptic_segmentation` (v5). Args: segmentation_array: Segmentatio

Process the result of Transformers segmentation functions such as `post_process_semantic_segmentation`, `post_process_instance_segmentation`,

Process the result of Transformers segmentation functions such as `post_process_semantic_segmentation` and `post_process_instance_segmentatio

Project state distribution to measurement space. Args: mean: The state's mean vector. covariance: The state'

Read a yaml file and return a dict. Args: file_path: The file path as a string or Path object. Returns: A dict of conte

Attempt to recover and parse a truncated or malformed JSON snippet from Qwen-2.5-VL output. This utility extracts a JSON-like portion fr

Reset the metric to its initial state, clearing all stored data.

Reset the metric to its initial state, clearing all stored data.

Reset the metric to its initial state, clearing all stored data.

Reset the counter to the initial start_id.

Write a dict to a json file. Args: data: dict with unique keys and value as pair. file_path: The file path as a string or Pa

Write the summary table to *path* as a CSV file. Each row mirrors the Rich summary table: scenario metadata, memory ratios, encode/decode ove

Write a list of strings to a text file, each string on a new line. Args: lines: The list of strings to be written to the file.

Save a dict to a yaml file. Args: data: dict with unique keys and value as pair. file_path: The file path as a string or Pat

Save dataset annotations in YOLO format. Args: dataset: The dataset whose annotations are saved. annotations_directory_path: Path

Scale the dimensions of bounding boxes. Args: xyxy: An array of shape `(n, 4)` containing the bounding boxes coordinates

Scale image by given factor. Scale factor > 1.0 zooms in, < 1.0 zooms out. Args: image: The image to scale. scale_factor: Fa

Time MaskAnnotator on both representations.

Time .area on both representations.

Synthesize polygon masks and build the CompactMask.

Time centroid: np.tensordot on full stack (dense) vs per-crop (compact).

Return (pixel_perfect, areas_match, roundtrip_ok).

Per-mask decode time: decode each mask individually and average over N. Building a list via compact_mask[i] decodes each crop separately, giving

Per-mask encode time: encode each mask individually and average over N. Calling from_dense one mask at a time (rather than batching all N) isolat

Time boolean filtering (keep every other detection).

Time pairwise self-IoU using dense (N,H,W) AND and compact crop filter. Correctness is checked on the first 10 masks only to keep it fast, re

Time Detections.merge on two half-splits. Dense: np.vstack; compact: RLE concat. Splits are pre-computed so the timed lambda measures only th

Time mask NMS. Dense resizes to 640 before IoU; compact uses exact crop IoU. Compact NMS is strictly more accurate than dense: it computes pixel-

Time mask offset: move_masks (N,H,W) copy vs O(N) offset update.

Time resize to half resolution; check pixel-level correctness. Dense path uses numpy fancy-indexing via ``_resize_dense_to_shape``. Compact p

Update a matched track. Args: new_track: The new track data. frame_id: The current frame ID.

Updates the tracker with the provided tensors and returns the updated tracks. Args: tensors: The new tensors to update w

Write and close the JSON file.

Converts bounding box coordinates from `(center_x, center_y, width, height)` format to `(x_min, y_min, x_max, y_max)` format. Args:

Converts a 2D `np.ndarray` of bounding boxes into a 3D `np.ndarray` of bool masks. Args: boxes: A 2D `np.ndarray` of shape `(N, 4)`

Convert an array of boxes to an array of polygons. Retains the input datatype. Args: box: An array of boxes (N, 4), where each b