Abstract: An electronic device includes a coding module that determines whether a parameter of an artificial neural network is an outlier, depending on a value of the parameter and compresses the parameter by truncating a first bit of the parameter when the parameter is a non-outlier and truncating a second bit of the parameter when the parameter is the outlier, and a decoding module that decodes a compressed parameter.

Abstract: An electronic device includes a coding module that determines whether a parameter of an artificial neural network is an outlier, depending on a value of the parameter and compresses the parameter by truncating a first bit of the parameter when the parameter is a non-outlier and truncating a second bit of the parameter when the parameter is the outlier, and a decoding module that decodes a compressed parameter.

Abstract: Black level calibration methods and systems are generally disclosed. According to one embodiment of the present invention, a method of calibrating a black level signal in a frame includes performing an iteration of averaging a first set of digital values corresponding to a first set of adjusted black level signals associated with a first set of black pixels of the frame, determining whether an average value based on the first set of digital values has reached a target black level, determining a calibration offset based on a difference between the average value and the target black level and an accumulator step, converting the calibration offset to an analog signal, generating a calibration signal based on the analog signal for a second set of black pixels of the frame, and repeating the iteration for the frame until a predetermined condition is determined to have been met.

Abstract: Multi-phase black level calibration (BLC) methods and systems are generally disclosed. According to one embodiment of the present invention, an image sensor comprises a pixel sensor array, a timing generator, and a front-end processing block. The front-end processing block also includes a first summing junction, a first BLC block, and a second BLC block. According to a first timing signal from the timing generator, the first BLC block is configured to iteratively generate a first calibration signal in a first phase based on a first set of adjusted black level signals associated with a first set of black pixels, a changing accumulator step, and a predetermined condition associated with a first target black level. According to a second timing signal from the timing generator, the second BLC block is configured to generate a second calibration signal for a second summing junction to apply to an image signal associated with one or more active pixels in the frame in a second phase.

Abstract: Multi-phase black level calibration (BLC) methods and systems are generally disclosed. According to one embodiment of the present invention, an image sensor comprises a pixel sensor array, a timing generator, and a front-end processing block. The front-end processing block also includes a first summing junction, a first BLC block, and a second BLC block. According to a first timing signal from the timing generator, the first BLC block is configured to iteratively generate a first calibration signal in a first phase based on a first set of adjusted black level signals associated with a first set of black pixels, a changing accumulator step, and a predetermined condition associated with a first target black level. According to a second timing signal from the timing generator, the second BLC block is configured to generate a second calibration signal for a second summing junction to apply to an image signal associated with one or more active pixels in the frame in a second phase.

Abstract: Black level calibration methods and systems are generally disclosed. According to one embodiment of the present invention, a method of calibrating a black level signal in a frame includes performing an iteration of averaging a first set of digital values corresponding to a first set of adjusted black level signals associated with a first set of black pixels of the frame, determining whether an average value based on the first set of digital values has reached a target black level, determining a calibration offset based on a difference between the average value and the target black level and an accumulator step, converting the calibration offset to an analog signal, generating a calibration signal based on the analog signal for a second set of black pixels of the frame, and repeating the iteration for the frame until a predetermined condition is determined to have been met.