Gopro, Inc.

Tom Médioni

Vincent Garcia

Francescu Santoni

Gabriel Lema

Matthieu Rouif

Video information defining video content may be obtained. The video content may include video frames and may have a progress length. The video frames may be encoded into video packets, with the video packets being of particular sizes. One or more size criteria for detecting a given moment within the video content may be obtained. The sizes of the video packets may be compared with the one or more size criteria. One or more sets of the video packets that satisfy the one or more size criteria may be identified. One or more portions of the video content having video frames defined by the set(s) of video packets that satisfy the one or more size criteria may be identified as the given moment within the video content. Storage of the identification of the given moment within the video content in a storage medium may be effectuated.

Systems and methods for detecting moments within videos

Convolutional neural network information may define a convolutional neural network including layers. The layers may define operations on an input to the convolutional neural network. The layers in the convolutional neural network information may be formatted as shaders. Input information defining the input to the convolutional neural network may be accessed. The input information may be formatted as an array of textures. The shaders may be applied to the textures to effectuate processing the input to the convolutional neural network through the layers of the convolutional neural network. One or more results may be obtained from applying the shaders to the array of textures.

Systems and methods for processing convolutional neural network operations using textures

Audio information defining audio content may be accessed. The audio content may have a duration. The audio content may be segmented into audio segments. Individual audio segments may correspond to a portion of the duration. The audio segments may include a first audio segment corresponding to a first portion of the duration. Energy features, entropy features, frequency features, and/or other features of the audio segments may be determined. Energy features may characterize energy of the audio segments. Entropy features may characterize spectral flatness of the audio segments. Frequency features may characterize highest frequencies of the audio segments. One or more of the audio segments may be identified as containing speech based on the energy features, the entropy features, the frequency features, and/or other information. Storage of the identification of the one or more of the audio segments as containing speech in one or more storage media may be effectuated.

Systems and methods for identifying speech based on spectral features

Audio content may have a duration. The audio content may be segmented into audio segments. Individual audio segments may correspond to a portion of the duration. Mel frequency spectral power features, Mel frequency cepstral coefficient features, and energy features of the audio segments may be determined. Feature vectors of the audio segments may be determined based on the Mel frequency spectral power features, the Mel frequency cepstral coefficient features, and the energy features. The feature vectors may be processed through a support vector machine. The support vector machine may output predictions on whether the audio segments contain speech. One or more of the audio segments may be identified as containing speech based on filtering the predictions and comparing the filtered predictions to a threshold. Storage of the identification of the one or more of the audio segments as containing speech in one or more storage media may be effectuated.

Systems and methods for identifying speech based on cepstral coefficients and support vector machines

Systems and method of automatically detecting user responses from users to media content and correlating the user responses to the media content include a determination of an emotional response by a user&#x2014;for example a smile&#x2014;to a presentation of media content. The determined emotional response is correlated to the media content. Subsequent presentation of the same media content includes presentation of an indicator&#x2014;for example a smiley&#x2014;that indicates the determined emotional response.

Systems and methods to detect and correlate user responses to media content

Video information defining video content may be accessed. The video content may include video frames. Motion vectors for the video frames may be determined. The motion vectors may represent motion of one or more visuals captured within individual video frames. A transformation matrix for the video frames may be determined based on the motion vectors. The transformation matrix may characterize rigid transformations between pairs of the video frames. Shakiness metrics for the video frames may be determined based on the transformation matrix. A shakiness threshold may be obtained. One or more of the video frames may be identified based on the shakiness metrics, the shakiness threshold, and algorithms with hysteresis or finite-state machines. A video summary of the video content may be generated. The video summary may include the one or more identified video frames.

Systems and methods for editing videos based on shakiness measures

A three-dimensional convolutional neural network may include a preliminary layer group, one or more intermediate layer groups, a final layer group, and/or other layers/layer groups. The preliminary layer group may include an input layer, a preliminary three-dimensional padding layer, a preliminary three-dimensional convolution layer, a preliminary activation layer, a preliminary normalization layer, and a preliminary downsampling layer. One or more intermediate layer groups may include an intermediate three-dimensional squeeze layer, a first intermediate normalization layer, an intermediate three-dimensional padding layer, a first intermediate three-dimensional expand layer, a second intermediate three-dimensional expand layer, an intermediate concatenation layer, a second intermediate normalization layer, an intermediate activation layer, and an intermediate combination layer. The final layer group may include a final dropout layer, a final three-dimensional convolution layer, a final activation layer, a final normalization layer, a final three-dimensional downsampling layer, and a final flatten layer.

Growing community of inventors

Paris, France

Tom Médioni