@brief Apply sigmoid activation: 1 / (1 + exp(-x)) @param x Input value @return Sigmoid of x

@brief Constructor - loads and initializes the ONNX model @param modelPath Path to the ONNX model file @param useGPU Whether to use GPU (CUDA) for inf

@brief Constructor @param modelPath Path to the ONNX model file @param labelsPath Path to the class names file @param useGPU Whether to use GPU for in

@brief Constructor @param modelPath Path to the ONNX model file @param labelsPath Path to the class names file @param useGPU Whether to use GPU for in

@brief Constructor @param modelPath Path to the ONNX model file @param labelsPath Path to the class names file @param useGPU Whether to use GPU for in

@brief Constructor @param modelPath Path to the ONNX model file @param labelsPath Path to the class names file (optional for pose) @param useGPU Wheth

@brief Constructor @param modelPath Path to the ONNX model file @param labelsPath Path to the class names file @param useGPU Whether to use GPU for in

@brief Compute area of the oriented bounding box

@brief Create a classifier with explicit version selection @param modelPath Path to the ONNX model @param labelsPath Path to the class names file @par

@brief Create a detector with explicit version selection @param modelPath Path to the ONNX model @param labelsPath Path to the class names file @param

@brief Create an input tensor from a blob @param blob Pointer to the input data @param inputTensorShape Shape of the input tensor @return ONNX Runtime

@brief Fast coordinate descaling (batch operation) @param coords Array of x,y coordinates to descale @param count Number of coordinate pairs @param sc

@brief Run pose detection on an image (optimized with buffer reuse) @param image Input image (BGR format) @param confThreshold Confidence threshold @p

@brief Run detection on an image (optimized with buffer reuse) @param image Input image (BGR format) @param confThreshold Confidence threshold @param

@brief Detect YOLO version for classification model @param outputShape The shape of the output tensor @return Detected YOLOVersion (V11 or V12 for cla

@brief Detect YOLO version from output tensors

@brief Draw detections on an image @param image Image to draw on @param detections Vector of detections

@brief Draw only segmentation masks (no boxes)

@brief Draw only skeletons (no bounding boxes)

@brief Get the input image shape expected by the model

@brief Get input shape

Classification doesn't track this

@brief Get letterbox padding and scale parameters @param originalShape Original image size @param letterboxShape Letterboxed image size @param[out] sc

@brief Get the number of input nodes

@brief Get the number of output nodes