Disclosed here are methods, systems, and devices for measuring visual quality degradation of digital content caused by an encoding process. There is received first data for a digital content item, which is not encoded by the encoding process, and second data for the digital content item, which is encoded by the encoding process. For a given artefact type, the first data and the second data are processed to obtain a first quality metric measuring visual quality degradation in the digital content item attributable to the given artefact type caused by the encoding process. A stored mapping corresponding to the given artefact type is applied to the first quality metric to obtain a second quality metric which measures visual quality degradation in the digital content item attributable to the given artefact type caused by the encoding process and approximates subjective assessment of the digital content item by a human visual system.

An image coding apparatus obtains a quantization parameter of a macroblock to be encoded. The quantization parameter is corrected by adding a correction value thereto. An encoding part encodes the macroblock by using the corrected quantization parameter. After the encoding, a quantization parameter correction part calculates the cumulative target amount of codes by accumulating the target amounts of codes set for the encoded macroblocks, respectively, and calculates the cumulative amount of generated codes by accumulating the respective amounts of generated codes of the encoded macroblocks. If the cumulative amount of generated codes is larger than the cumulative target amount of codes, the quantization parameter correction part increments the correction value. A new macroblock to be encoded is quantized more coarsely than the encoded macroblocks.

An image coding apparatus obtains a quantization parameter of a macroblock to be encoded. The quantization parameter is corrected by adding a correction value thereto. An encoding part encodes the macroblock by using the corrected quantization parameter. After the encoding, a quantization parameter correction part calculates the cumulative target amount of codes by accumulating the target amounts of codes set for the encoded macroblocks, respectively, and calculates the cumulative amount of generated codes by accumulating the respective amounts of generated codes of the encoded macroblocks. If the cumulative amount of generated codes is larger than the cumulative target amount of codes, the quantization parameter correction part increments the correction value. A new macroblock to be encoded is quantized more coarsely than the encoded macroblocks.

Non-limiting examples of the present disclosure describe detection of gross motion of a region of content. Gross motion of a region of content may be detected. A determination may be made as to a current quality level of the region. Based on detection of the gross motion, residual values may be generated for a progressive update of the region. The residual values are generated using the current quality level of the region as a base to determine a quantization update for a progressive update of the region at a higher quality level as compared with the current quality level of the region. Frame data for the progressive update of the region may be encoded. The frame data may comprise the residual values and motion vectors for progressive update of the region. The frame data may be transmitted for decoding. Other examples are also described.

Techniques related to improved video coding based on face detection, region extraction, and tracking are discussed. Such techniques may include performing a facial search of a video frame to determine candidate face regions in the video frame, testing the candidate face regions based on skin tone information to determine valid and invalid face regions, rejecting invalid face regions, and encoding the video frame based on valid face regions to generate a coded bitstream.

Machine learning is used to refine a probability distribution for entropy coding video or image data. A probability distribution is determined for symbols associated with a video block (e.g., quantized transform coefficients, such as during encoding, or syntax elements from a bitstream, such as during decoding), and a set of features is extracted from video data associated with the video block and/or neighbor blocks. The probability distribution and the set of features are then processed using machine learning to produce a refined probability distribution. The video data associated with a video block are entropy coded according to the refined probability distribution. Using machine learning to refine the probability distribution for entropy coding minimizes the cross-entropy loss between the symbols to entropy code and the refined probability distribution.

A method of multi-rate neural image compression is performed by at least one processor and includes selecting encoding masks, based on a first hyperparameter, and performing a convolution of a first plurality of weights of a first neural network and the selected encoding masks to obtain first masked weights. The method further includes encoding an input image to obtain an encoded representation, using the first masked weights, and encoding the obtained encoded representation to obtain a compressed representation.

An example video encoder is configured to receive an indication of merge mode coding of a block within a parallel motion estimation region (PMER), generate a merge mode candidate list comprising one or more spatial neighbor motion vector (MV) candidates and one or more temporal motion vector prediction (TMVP) candidates, wherein motion information of at least one of the spatial neighbor MV candidates is known to be unavailable during coding of the block at an encoder, determine an index value identifying, within the merge mode candidate list, one of the TMVP candidates or the spatial neighbor MV candidates for which motion information is available during coding of the particular block, and merge mode code the block using the identified MV candidate.

An example video encoder is configured to receive an indication of merge mode coding of a block within a parallel motion estimation region (PMER), generate a merge mode candidate list comprising one or more spatial neighbor motion vector (MV) candidates and one or more temporal motion vector prediction (TMVP) candidates, wherein motion information of at least one of the spatial neighbor MV candidates is known to be unavailable during coding of the block at an encoder, determine an index value identifying, within the merge mode candidate list, one of the TMVP candidates or the spatial neighbor MV candidates for which motion information is available during coding of the particular block, and merge mode code the block using the identified MV candidate.

The present invention generally relates to a method for predicting a block of pixels from at least one patch comprising a block of pixels and a causal neighborhood around this block of pixels. The method comprises the following steps: determining a mapping of a causal neighborhood, around that block of pixels to be predicted, on the block of pixels to be predicted in order that the block of pixels of each patch is best predicted by mapping the neighborhood of that patch on the block of pixels of that patch, and predicting the block of pixels from a prediction block computed by applying the determined mapping on the neighborhood of the block of pixels to predict.