Abstract: Methods, mediums, and systems for constructing a 3D face model, including Deep Iterative Face Fitting (DIFF) and Recurrent Feature Alignment (ReFA). A processor provides a reference—containing a 3D mesh representing a median face, vertex positions, or surface points—as a template and UV texture mapping pointing the 3D mesh to a 2D UV space; receives input image(s) of a face and extracts geometry/texture features in an image space; extracts features in a UV space; iteratively produces a feature map via visual semantic correlation between UV and image spaces and regress geometry updates, predicting texture maps and comparing features, and inputs the map to an RNN-based neural optimizer of Gated Recurrent Units (GRU) to determine a hidden state. A head pose and/or updated 3D mesh/UV-space position map is output, each pixel in the UV-space map storing a coordinate of a corresponding point in a canonical space of the 3D mesh.

Abstract: A method and device for switching video pictures are disclosed in the invention, which belong to video conference field. Said method includes: obtaining correspondence relations between participants and venues; finding, when a participant is called, a venue corresponding to the called participant according to the correspondence relations; and switching the video pictures to the venue corresponding to the called participant. According to the invention, it is possible to find the venue corresponding to the called participant according to the correspondence relations, and automatically switch the video to the venue or request to an MCU for switching the video to the venue, without a manual intervention of the participant or waiting for a response from the participant, so that continuity of the video conference is ensured, the video pictures can be switched in time, and “face to face” communion experience of the participants in the video conference is improved.

Abstract: Embodiments of the present invention disclose an audio encoding method. The method includes: obtaining audio data of N channels; and performing channel interleaving and packetization on the obtained audio data of the N channels to obtain data packets, where each data packet includes X*N segments of audio data, where X is a ratio of an amount of audio data included in one data packet to an amount of audio data included in one audio frame, X is an integer greater than or equal to 1, and in the X*N segments of audio data, at least X+1 segments of audio data belong to different audio frames.

Abstract: An audio decoding method and apparatus are disclosed. The audio decoding method includes: receiving data packets; when data packet loss is detected and audio data of an audio frame corresponding to M channels in N channels is lost, if audio data of other channels than the M channels in the N channels, which belongs to the same audio frame as the lost audio data in the audio frame, is not lost, decoding the un-lost audio data; extracting a signal characteristic parameter of the data obtained after decoding; determining whether a correlation exists between a first channel and a second channel; and if the correlation exists, performing packet loss concealment processing on the lost audio data of the audio frame corresponding to the first channel according to the second channel. The audio decoding method and apparatus can effectively improve the effect of packet loss concealment processing in audio decoding.

Abstract: A method and device for switching video pictures are disclosed in the invention, which belong to video conference field. Said method includes: obtaining correspondence relations between participants and venues; finding, when a participant is called, a venue corresponding to the called participant according to the correspondence relations; and switching the video pictures to the venue corresponding to the called participant. According to the invention, it is possible to find the venue corresponding to the called participant according to the correspondence relations, and automatically switch the video to the venue or request to an MCU for switching the video to the venue, without a manual intervention of the participant or waiting for a response from the participant, so that continuity of the video conference is ensured, the video pictures can be switched in time, and “face to face” communion experience of the participants in the video conference is improved.