Efficient channel swapping

[MatthijsBurgh]

Motivation

Swapping the channels manually is much faster than the conversion function. It is also independent of the image size.

In [1]: import numpy as np

In [2]: import cv2

In [3]: bla = (np.random.rand(1080, 1920, 3) * 255).astype(np.float32)

In [4]: bla2 = (np.random.rand(2080, 3920, 3) * 255).astype(np.float32)

In [5]: timeit cv2.cvtColor(bla, cv2.COLOR_BGR2RGB)
818 μs ± 14.3 μs per loop (mean ± std. dev. of 7 runs, 1,000 loops each)

In [6]: timeit cv2.cvtColor(bla2, cv2.COLOR_BGR2RGB)
11.7 ms ± 79.6 μs per loop (mean ± std. dev. of 7 runs, 100 loops each)

In [7]: timeit bla[..., ::-1]
77.7 ns ± 3.16 ns per loop (mean ± std. dev. of 7 runs, 10,000,000 loops each)

In [8]: timeit bla2[..., ::-1]
75.7 ns ± 0.438 ns per loop (mean ± std. dev. of 7 runs, 10,000,000 loops each)

Modification

Replace cvtColor calls to manually swapping for BGR2RGB or vice-versa.

BC-breaking (Optional)

Nope

Use cases (Optional)

No new features

Checklist

Before PR:

• I have read and followed the workflow indicated in the CONTRIBUTING.md to create this PR.
• Pre-commit or linting tools indicated in CONTRIBUTING.md are used to fix the potential lint issues.
• Bug fixes are covered by unit tests, the case that causes the bug should be added in the unit tests.
• New functionalities are covered by complete unit tests. If not, please add more unit test to ensure the correctness.
• The documentation has been modified accordingly, including docstring or example tutorials.

After PR:

• If the modification has potential influence on downstream or other related projects, this PR should be tested with some of those projects, like MMDet or MMCls.
• CLA has been signed and all committers have signed the CLA in this PR.

[CLAassistant]

All committers have signed the CLA.