COCO Processing API

API reference for the COCO Processing module.

COCOProcessor

Main class for processing COCO annotation files.

Constructor

COCOProcessor(annotation_file: str, image_dir: str = None)

Parameters:

• annotation_file (str): Path to COCO annotation JSON file
• image_dir (str, optional): Directory containing images

Example:

processor = COCOProcessor('annotations.json', 'images/')

Methods

load_annotations()

Load and parse COCO annotations.

def load_annotations() -> dict

Returns: Dictionary containing parsed annotations

---

extract_masks()

Extract instance segmentation masks.

def extract_masks(
    category: str = None,
    image_id: int = None
) -> dict

Parameters:

• category (str, optional): Filter by category name
• image_id (int, optional): Filter by image ID

Returns: Dictionary mapping instance IDs to binary masks

---

get_categories()

Get list of all categories in the dataset.

def get_categories() -> list

Returns: List of category dictionaries

---

filter_by_area()

Filter annotations by area.

def filter_by_area(
    min_area: float = 0,
    max_area: float = float('inf')
) -> list

Parameters:

• min_area (float): Minimum area threshold
• max_area (float): Maximum area threshold

Returns: Filtered list of annotations

---

visualize_annotations()

Visualize annotations on image.

def visualize_annotations(
    image_id: int,
    save_path: str = None,
    show: bool = True
) -> np.ndarray

Parameters:

• image_id (int): Image ID to visualize
• save_path (str, optional): Path to save visualization
• show (bool): Whether to display the image

Returns: Annotated image as numpy array