skvideo.measure
provides quality assessment tools, scene detection, and other measurement operations.
Use skvideo.measure.strred
to measure the perceptually quality difference between two videos, providing a slightly reduced reference score along with a critically reduced reference score (1 number comparison between videos).
Use skvideo.measure.ssim
to measure the perceptually quality difference between two videos, considering only individual frames.
Use skvideo.measure.msssim
to measure the perceptually quality difference between two videos, considering only individual frames. Differs from ssim in that this function considers multiple scales.
Use skvideo.measure.mse
, skvideo.measure.mad
, or skvideo.measure.psnr
to measure simple point-wise similarity between two videos.
Use skvideo.measure.brisque_features
to extract frame-by-frame features using the BRISQUE image quality algorithm, useful for frame-level quality analysis and training your own frame-based quality-aware model.
Use skvideo.measure.videobliinds_features
to extract features using the Video Bliinds algorithm, useful for training your own quality-aware model.
Use skvideo.measure.viideo_features
to extract features using the Viideo Oracle algorithm, useful for training your own quality-aware model.
Use skvideo.measure.viideo_score
to return the score of the Viideo Oracle algorithm.
Use skvideo.measure.niqe
to return per-frame Naturalness Image Quality Evaluator (NIQE) scores.
Use skvideo.measure.scenedet
to find frames that start scene boundaries. Check the documentation on selecting the specific detection algorithm.