Lighting

class mmpretrain.datasets.transforms.Lighting(eigval, eigvec, alphastd=0.1, to_rgb=False)

Adjust images lighting using AlexNet-style PCA jitter.

Required Keys:

• img

Modified Keys:

• img

Parameters:

• eigval (Sequence[float]) – the eigenvalue of the convariance matrix of pixel values, respectively.

• eigvec (list[list]) – the eigenvector of the convariance matrix of pixel values, respectively.

• alphastd (float) – The standard deviation for distribution of alpha. Defaults to 0.1.

• to_rgb (bool) – Whether to convert img to rgb. Defaults to False.

transform(results)

Transform function to resize images.

Parameters:

results (dict) – Result dict from loading pipeline.

Returns:

Lightinged results, ‘img’ key is updated in result dict.

Return type:

dict