Module preprocessing 

Image preprocessing for YOLO inference.

This module handles all image preprocessing operations needed before running YOLO model inference, including resizing, padding, and normalization.

Structs

PreprocessResult

Result of preprocessing an image, containing the tensor and transform info.

Constants

LETTERBOX_COLOR

Default letterbox padding color (gray).

Functions

array_to_tensor

Convert a raw HWC u8 array to a normalized NCHW tensor.

calculate_rect_size

Calculate target size for rectangular inference mode.

clip_coords

Clip coordinates to image bounds.

image_to_array

Convert a DynamicImage to an HWC ndarray.

preprocess_image

Preprocess an image for YOLO inference.

preprocess_image_center_crop

Preprocess an image for YOLO classification (Center Crop).

preprocess_image_with_precision

Preprocess an image for YOLO inference with optional FP16 output.

scale_coords

Scale coordinates from model output space back to original image space.