[Feature]: Kornia-rs still largely slower than OpenCV in 2026

🚀 Feature Description

tested simple

import time
import cv2
import numpy as np
import kornia_rs as K

cv2.setNumThreads(1)

sizes = [
    (256, 256),
    (512, 512),
    (1024, 1024),
    (2048, 2048),
]

img = cv2.imread("data/images/test_ocr_page2.png", cv2.IMREAD_COLOR)
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

iters = 200
warmup = 20


def bench_opencv(img, size):
    h, w = size
    for _ in range(warmup):
        cv2.resize(img, (w, h), interpolation=cv2.INTER_LINEAR)

    t0 = time.perf_counter()
    for _ in range(iters):
        cv2.resize(img, (w, h), interpolation=cv2.INTER_LINEAR)
    return (time.perf_counter() - t0) / iters


def bench_kornia(img, size):
    h, w = size
    for _ in range(warmup):
        K.resize(img, (h, w), interpolation="bilinear")

    t0 = time.perf_counter()
    for _ in range(iters):
        K.resize(img, (h, w), interpolation="bilinear")
    return (time.perf_counter() - t0) / iters


print(f"Iterations: {iters}, Warmup: {warmup}")
print("-" * 60)

for size in sizes:
    t_cv = bench_opencv(img, size)
    t_kr = bench_kornia(img, size)

    print(
        f"{size[0]:4d}x{size[1]:4d} | "
        f"OpenCV: {t_cv*1000:6.2f} ms | "
        f"kornia-rs: {t_kr*1000:6.2f} ms | "
        f"ratio: {t_kr / t_cv:5.2f}x"
    )

📂 Feature Category

Rust Core Library

💡 Motivation

I found the result is surprisely badly lay behind OpenCV tooo much:

Iterations: 200, Warmup: 20
------------------------------------------------------------
 256x 256 | OpenCV:   0.16 ms | kornia-rs:   0.80 ms | ratio:  4.93x
 512x 512 | OpenCV:   0.62 ms | kornia-rs:   1.26 ms | ratio:  2.04x
1024x1024 | OpenCV:   1.78 ms | kornia-rs:   3.03 ms | ratio:  1.71x
2048x2048 | OpenCV:   4.47 ms | kornia-rs:   8.34 ms | ratio:  1.87x

💭 Proposed Solution

Wondering if kornia-rs pay some time on improve simple operation such as resize rather than spent too much converting kornia to 3D libary.

📚 Library Reference

opencv

🔄 Alternatives Considered

No response

🎯 Use Cases

No response

📝 Additional Context

No response

🤝 Contribution Intent

• I plan to submit a PR to implement this feature
• I'm requesting this feature but not planning to implement it