Implementation:Kornia Kornia ONNXSequential Call

Knowledge Sources	Kornia
Domains	ONNX, Deployment, Inference
Last Updated	2026-02-09 15:00 GMT

Overview

Concrete tool for running ONNX sequential pipeline inference provided by Kornia's ONNXRuntimeMixin.

Description

ONNXSequential inherits __call__ from ONNXRuntimeMixin. The method accepts numpy arrays as positional arguments, maps them to the model's input names, executes the ONNX Runtime session, and returns a list of numpy output arrays. It handles input validation and output collection.

Usage

Call the ONNXSequential instance directly with numpy input arrays.

Code Reference

Repository	https://github.com/kornia/kornia
File	`kornia/core/mixin/onnx.py`
Lines	L274–288
Signature	`def __call__(self, *inputs: np.ndarray) -> list[np.ndarray]`
Import	`from kornia.onnx import ONNXSequential` (call via instance)

I/O Contract

Inputs

Parameter	Type	Required	Description
`*inputs`	`np.ndarray`	Yes	Numpy arrays matching model input shapes and dtypes

Outputs

list[np.ndarray] — model outputs as numpy arrays.

Usage Examples

Running inference on preprocessed numpy data

import numpy as np
from kornia.onnx import ONNXSequential

# Construct pipeline
pipeline = ONNXSequential(
    "hf://operators/resize",
    "hf://operators/classifier",
    io_maps=[("resized", "input_image")],
)

# Prepare input as numpy array (batch, channels, height, width)
input_data = np.random.randn(1, 3, 224, 224).astype(np.float32)

# Run inference
outputs = pipeline(input_data)
print(outputs[0].shape)  # e.g., (1, 1000)

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment