Abstract: A computer system generates augmented training datasets to train neural network models. The computer system receives an initial training dataset comprising images for training a neural network model, and generates an augmented training dataset by modifying images from the first training dataset. The computer system identifies a representation of a target object against a background from the initial training dataset and extracts a portion of the image displaying the target object. The computer system generates samples for including in the augmented training dataset based on the image. For example, new images may be obtained by performing transformations on the portion of the image displaying the target object and/or by overlaying the transformed portion of the image over a different background. The modified images are included in the augmented training dataset used for training the neural network model to recognize the target object.

Abstract: A computer system generates augmented training datasets to train neural network models. The computer system receives an initial training dataset comprising images for training a neural network model, and generates an augmented training dataset by modifying images from the first training dataset. The computer system identifies a representation of a target object against a background from the initial training dataset and extracts a portion of the image displaying the target object. The computer system generates samples for including in the augmented training dataset based on the image. For example, new images may be obtained by performing transformations on the portion of the image displaying the target object and/or by overlaying the transformed portion of the image over a different background. The modified images are included in the augmented training dataset used for training the neural network model to recognize the target object.

Abstract: A computer system generates augmented training datasets to train neural network models. The computer system receives an initial training dataset comprising images for training a neural network model, and generates an augmented training dataset by modifying images from the first training dataset. The computer system identifies a representation of a target object against a background from the initial training dataset and extracts a portion of the image displaying the target object. The computer system generates samples for including in the augmented training dataset based on the image. For example, new images may be obtained by performing transformations on the portion of the image displaying the target object and/or by overlaying the transformed portion of the image over a different background. The modified images are included in the augmented training dataset used for training the neural network model to recognize the target object.

Abstract: A computer system generates augmented training datasets to train neural network models. The computer system receives an initial training dataset comprising images for training a neural network model, and generates an augmented training dataset by modifying images from the first training dataset. The computer system identifies a representation of a target object against a background from the initial training dataset and extracts a portion of the image displaying the target object. The computer system generates samples for including in the augmented training dataset based on the image. For example, new images may be obtained by performing transformations on the portion of the image displaying the target object and/or by overlaying the transformed portion of the image over a different background. The modified images are included in the augmented training dataset used for training the neural network model to recognize the target object.

Abstract: A flexible timebase for eye diagrams uses a stable free running oscillator as a sample clock for equivalent time sampling of an input serial digital signal and of a reference signal, derived from a subdivided recovered clock of the input serial digital signal. The reference signal samples are provided to a digital phase-locked loop that provides the flexible timebase to an eye pattern generator. The eye pattern generator accumulates the input serial digital signal samples at times corresponding to the reference signal samples to produce the eye diagram. A linear phase detector in the digital phase locked loop converts the reference signal samples to a complex signal using a Hilbert transform and then to a linear ramp of phase values using a CORDIC algorithm with arctangent lookup table. A subtractor then subtracts the digital phase-locked loop feedback from the linear ramp to provide the input to the loop filter.

Abstract: A flexible timebase for eye diagrams uses a stable free running oscillator as a sample clock for equivalent time sampling of an input serial digital signal and of a reference signal derived from a subdivided recovered clock of the input serial digital signal. The reference signal samples are provided to a digital phase-locked loop that provides the flexible timebase to an eye pattern generator. The eye pattern generator accumulates the input serial digital signal samples at times corresponding to the reference signal samples to produce the eye diagram. A linear phase detector in the digital phase locked loop converts the reference signal samples to a complex signal using a Hilbert transform and then to a linear ramp of phase values using a CORDIC algorithm with arctangent lookup table. The digital phase-locked loop feedback is subtracted from the linear ramp to provide the input to the loop filter.

Abstract: A flexible timebase for eye diagrams uses a stable free running oscillator as a sample clock for equivalent time sampling of an input serial digital signal and of a reference signal, such as a sine wave, derived from a subdivided recovered clock of the input serial digital signal. The reference signal samples are provided to a digital phase-locked loop that provides the flexible timebase to an eye pattern generator. The eye pattern generator accumulates the input serial digital signal samples at times corresponding to the reference signal samples to produce the eye diagram. A linear phase detector in the digital phase locked loop converts the reference signal samples to a complex signal using a Hilbert transform and then to a linear ramp of phase values using a CORDIC algorithm with arctangent lookup table. A subtractor is then used to subtract the digital phase-locked loop feedback from the linear ramp to provide the input to the loop filter.

Abstract: Image alias rejection when converting a high resolution rasterized waveform to a lower resolution rasterized waveform for display uses a statistical filter. The statistical filter provides a shaped probability density function either by combining the outputs of multiple random number generators, such as linear feedback shift registers, or by using a corresponding look-up table to produce a dither signal. The statistical filter may be applied to one or both of the dimensional values for each data point of the high resolution rasterized waveform by combining the dimensional values with the dither signal. The resulting filtered dimensional values may then be subsampled, such as by truncation, to produce values for a lower resolution rasterized waveform display.

Abstract: A flexible timebase for eye diagrams uses a stable free running oscillator as a sample clock for equivalent time sampling of an input serial digital signal and of a reference signal, such as a sine wave, derived from a subdivided recovered clock of the input serial digital signal. The reference signal samples are provided to a digital phase-locked loop that provides the flexible timebase to an eye pattern generator. The eye pattern generator accumulates the input serial digital signal samples at times corresponding to the reference signal samples to produce the eye diagram. A linear phase detector in the digital phase locked loop converts the reference signal samples to a complex signal using a Hilbert transform and then to a linear ramp of phase values using a CORDIC algorithm with arctangent lookup table. A subtractor is then used to subtract the digital phase-locked loop feedback from the linear ramp to provide the input to the loop filter.

Abstract: A video field rate persistence is provided for a video waveform display where multiple video components are displayed simultaneously, one component being updated per field interval. A persistence count is assigned to each pixel in a circular field buffer which is divided equally into a number of sub-buffers representing a portion of a display raster, one for each component. The persistence count is updated for each update of the circular field buffer with the pixel values being maintained until the persistence count equals a maximum value, at which point the pixel values for that component are updated with new pixel values. In this manner the display pixels for each component are updated every X video fields and are maintained at their prior values between updates to provide a flicker free display without temporal separation.

Abstract: Audio processing using a video rasterizer allows simultaneous display of both audio and video data on the same raster display. Audio samples are clocked into a buffer memory, such as a first-in, first-out (FIFO) buffer, by an audio clock and clocked out of the buffer memory by a rasterizer (system) clock to allow the audio data to cross over from the audio clock domain to the video clock domain. The audio data is then upsampled by a sample ratio converter/interpolator before being processed by a video display engine. The video display engine includes a polyphase filter for upsampling display data for input to a rasterizer from which the data is read for display.

Abstract: A video field rate persistence is provided for a video waveform display where multiple video components are displayed simultaneously, one component being updated per field interval. A persistence count is assigned to each pixel in a circular field buffer which is divided equally into a number of sub-buffers representing a portion of a display raster, one for each component. The persistence count is updated for each update of the circular field buffer with the pixel values being maintained until the persistence count equals a maximum value, at which point the pixel values for that component are updated with new pixel values. In this manner the display pixels for each component are updated every X video fields and are maintained at their prior values between updates to provide a flicker free display without temporal separation.

Abstract: Audio processing using a video rasterizer allows simultaneous display of both audio and video data on the same raster display. Audio samples are clocked into a buffer memory, such as a first-in, first-out (FIFO) buffer, by an audio clock and clocked out of the buffer memory by a rasterizer (system) clock to allow the audio data to cross over from the audio clock domain to the video clock domain. The audio data is then upsampled by a sample ratio converter/interpolator before being processed by a video display engine. The video display engine includes a polyphase filter for upsampling display data for input to a rasterizer from which the data is read for display.

Abstract: Image alias rejection when converting a high resolution rasterized waveform to a lower resolution rasterized waveform for display uses a statistical filter. The statistical filter provides a shaped probability density function either by combining the outputs of multiple random number generators, such as linear feedback shift registers, or by using a corresponding look-up table to produce a dither signal. The statistical filter may be applied to one or both of the dimensional values for each data point of the high resolution rasterized waveform by combining the dimensional values with the dither signal. The resulting filtered dimensional values may then be subsampled, such as by truncation, to produce values for a lower resolution rasterized waveform display.