View Source API Reference YOLO v0.1.2

Behaviour module defining callbacks for part of the image preprocessing required by YOLO models.

Provides functions for resizing images while preserving aspect ratio and handling padding to reach target dimensions.

FrameScaler for when the input image is already a tensor with shape {height, width, 3} and no scaling is needed because width and height already match the model input size. This is useful when calling YOLO.detect/3 with a tensor that already matches the expected dimensions.

Stores image scaling and padding configuration used during YOLO model preprocessing.

Defines a behaviour for implementing YOLO object detection models.

This module handles loading YOLO models and running object detection on images. The default implementation supports YOLOv8 models, but the YOLO.Model behaviour can be implemented for other YOLO variants.

Ultralytics model implementation for preprocessing input images and postprocessing detections using non-maximum suppression (NMS).

YoloV8 model implementation that delegates to the Ultralytics module. This module is deprecated and will be removed in the future.

Elixir NMS (Non-Maximum Suppression)

Utils to transform the model input and output.