Abstract: A self-tuning model compression methodology for reconfiguring a Deep Neural Network (DNN) includes: receiving a pre-trained DNN model and a data set; performing an inter-layer sparsity analysis to generate a first sparsity result; and performing an intra-layer sparsity analysis to generate a second sparsity result, including: defining a plurality of sparsity metrics for the network; performing forward and backward passes to collect data corresponding to the sparsity metrics; using the collected data to calculate values for the defined sparsity metrics; and visualizing the calculated values using at least a histogram.

Abstract: A Fast Non-Maximum Suppression (NMS) Algorithm post-processing for object detection includes getting original data output from a deep learning model inference output, the original data including a plurality of bounding boxes, pre-emptively filtering out at least one bounding box of the plurality of bounding boxes from further consideration when applying the algorithm, the at least one bounding box filtered out according to a predetermined criteria, processing data, using sigmoid functions or exponential functions, from bounding boxes of the plurality of bounding boxes not filtered out to generate processed bounding boxes, calculating final scores of the processed bounding boxes, and choosing a processed bounding boxes utilizing the final scores.

Abstract: A system for operating a floating-to-fixed arithmetic framework includes a floating-to-fix arithmetic framework on an arithmetic operating hardware such as a central processing unit (CPU) for computing a floating pre-trained convolution neural network (CNN) model to a dynamic fixed-point CNN model. The dynamic fixed-point CNN model is capable of implementing a high performance convolution neural network (CNN) on a resource limited embedded system such as mobile phone or video cameras.

Abstract: A method of pruning a batch normalization layer from a pre-trained deep neural network model is proposed. The pre-trained deep neural network model is inputted as a candidate model. The candidate model is pruned by removing the at least one batch normalization layer from the candidate model to form a pruned candidate model only when the at least one batch normalization layer is connected to and adjacent to a corresponding linear operation layer. The corresponding linear operation layer may be at least one of a convolution layer, a dense layer, a depthwise convolution layer, and a group convolution layer. Weights of the corresponding linear operation layer are adjusted to compensate for the removal of the at least one batch normalization. The pruned candidate model is then output and utilized for inference.

Abstract: A Fast Non-Maximum Suppression (NMS) Algorithm post-processing for object detection includes getting original data output from a deep learning model inference output, the original data including a plurality of bounding boxes, pre-emptively filtering out at least one bounding box of the plurality of bounding boxes from further consideration when applying the algorithm, the at least one bounding box filtered out according to a predetermined criteria, processing data, using sigmoid functions or exponential functions, from bounding boxes of the plurality of bounding boxes not filtered out to generate processed bounding boxes, calculating final scores of the processed bounding boxes, and choosing a processed bounding boxes utilizing the final scores.

Abstract: A method of compressing a pre-trained deep neural network model includes inputting the pre-trained deep neural network model as a candidate model. The candidate model is compressed by increasing sparsity of the candidate, removing at least one batch normalization layer present in the candidate model, and quantizing all remaining weights into fixed-point representation to form a compressed model. Accuracy of the compressed model is then determined utilizing an end-user training and validation data set. Compression of the candidate model is repeated when the accuracy improves. Hyper parameters for compressing the candidate model are adjusted, then compression of the candidate model is repeated when the accuracy declines. The compressed model is output for inference utilization when the accuracy meets or exceeds the end-user performance metric and target.

Abstract: A computing network includes a plurality of processing nodes. A method of training the computing network includes a processing node in the plurality of processing nodes computing an output estimate according to a weight defined by a weight variable and a connectivity mask, and adjusting connectivity variables according to an objective function to reduce a total number of connections between the plurality of processing nodes and reduce a performance loss indicative of how different the output estimate is from a target value. The connectivity mask represents a connection between the processing node and a preceding processing node in the plurality of processing nodes and is derived from a connectivity variable.

Abstract: An image processing system includes an image capturing device, a pixel binning device, a temporal filter, a first memory, a re-mosaic device, a second memory, and a blending device. The image capturing device is used for capturing a raw image. The pixel binning device is coupled to the image capturing device for outputting an enhanced image according to the raw image. The temporal filter is coupled to the pixel binning device for outputting a preview image according to the enhanced image. The first memory is used for buffering the raw image. The re-mosaic device is coupled to the first memory for outputting a processed image. The second memory is used for buffering the enhanced image. The blending device is coupled to the re-mosaic device and the second memory for outputting a snapshot image according to the processed image and the enhanced image.

Abstract: After inputting input data to a floating pre-trained convolution neural network to generate floating feature maps for each layer of the floating pre-trained CNN model, a statistical analysis on the floating feature maps is performed to generate a dynamic quantization range for each layer of the floating pre-trained CNN model. Based on the obtained quantization range for each layer, the proposed quantization methodologies quantize the floating pre-trained CNN model to generate the scalar factor of each layer and the fractional bit-width of a quantized CNN model. It enables the inference engine performs low-precision fixed-point arithmetic operations to generate a fixed-point inferred CNN model.

Abstract: A system for operating a floating-to-fixed arithmetic framework includes a floating-to-fix arithmetic framework on an arithmetic operating hardware such as a central processing unit (CPU) for computing a floating pre-trained convolution neural network (CNN) model to a dynamic fixed-point CNN model. The dynamic fixed-point CNN model is capable of implementing a high performance convolution neural network (CNN) on a resource limited embedded system such as mobile phone or video cameras.

Abstract: A face recognition module includes a near infrared flash, a master near infrared camera, an artificial intelligence NIR image model, an artificial intelligence original image model, and an artificial intelligence fusion model. The NIR flash flashes near infrared light. The master near infrared camera captures a NIR image. The artificial intelligence NIR image model processes the NIR image to generate NIR features. The artificial intelligence original image model processes a 2 dimensional second camera image to generate face features or color features. The artificial intelligence fusion model generates 3 dimensional face features, a depth map and an object's 3 dimensional model according to the NIR features, the face features and the color features.

Abstract: A self-tuning model compression methodology for reconfiguring a Deep Neural Network includes: receiving a DNN model and a data set, wherein the DNN includes an input layer, at least one hidden layer and an output layer, and said at least one hidden layer and the output layer of the DNN model includes a plurality of neurons; compressing the DNN model into a reconfigured model according to the data set, wherein the reconfigured model includes an input layer, at least one hidden layer and an output layer, and said at least one hidden layer and the output layer of the reconfigured model includes a plurality of neurons, and a size of the reconfigured model is smaller than a size of the DNN model; and executing the reconfigured model on a user terminal for an end-user application.

Abstract: A method of compressing a pre-trained deep neural network model includes inputting the pre-trained deep neural network model as a candidate model. The candidate model is compressed by increasing sparsity of the candidate, removing at least one batch normalization layer present in the candidate model, and quantizing all remaining weights into fixed-point representation to form a compressed model. Accuracy of the compressed model is then determined utilizing an end-user training and validation data set. Compression of the candidate model is repeated when the accuracy improves. Hyper parameters for compressing the candidate model are adjusted, then compression of the candidate model is repeated when the accuracy declines. The compressed model is output for inference utilization when the accuracy meets or exceeds the end-user performance metric and target.

Abstract: A method of pruning a batch normalization layer from a pre-trained deep neural network model is proposed. The pre-trained deep neural network model is inputted as a candidate model. The candidate model is pruned by removing the at least one batch normalization layer from the candidate model to form a pruned candidate model only when the at least one batch normalization layer is connected to and adjacent to a corresponding linear operation layer. The corresponding linear operation layer may be at least one of a convolution layer, a dense layer, a depthwise convolution layer, and a group convolution layer. Weights of the corresponding linear operation layer are adjusted to compensate for the removal of the at least one batch normalization. The pruned candidate model is then output and utilized for inference.

Abstract: System and methods are provided for touch detection. An example system includes: a measurement unit configured to acquire capacitance measurement data from a touch panel; a pre-processing unit configured to detect whether a touch event occurs on the touch panel based at least in part on the capacitance measurement data and generate an activation signal in response to the detection of the touch event; and a microcontroller unit configured to be activated in response to the activation signal to perform post-processing operations related to the touch event.

Abstract: System and methods are provided for tracking baseline signals for touch detection. The system includes: a comparison network configured to determine whether an input baseline signal is within a tracking range; a filter network configured to generate an output baseline signal for touch detection based at least in part on the input baseline signal according to one or more filter parameters; and a signal processing component configured to update the one or more filter parameters based at least in part on the determination of whether the input baseline signal is within the tracking range.

Abstract: System and methods are provided for touch detection. An example system includes: a measurement unit configured to acquire capacitance measurement data from a touch panel; a pre-processing unit configured to detect whether a touch event occurs on the touch panel based at least in part on the capacitance measurement data and generate an activation signal in response to the detection of the touch event; and a microcontroller unit configured to be activated in response to the activation signal to perform post-processing operations related to the touch event.

Abstract: System and methods are provided for tracking baseline signals for touch detection. The system includes: a comparison network configured to determine whether an input baseline signal is within a tracking range; a filter network configured to generate an output baseline signal for touch detection based at least in part on the input baseline signal according to one or more filter parameters; and a signal processing component configured to update the one or more filter parameters based at least in part on the determination of whether the input baseline signal is within the tracking range.

Abstract: This disclosure describes systems and techniques for implementing a touchscreen. These systems and/or techniques enable processing of a signal generated from one or more sensors of a touchscreen to reduce noise and increase accuracy.

Abstract: This disclosure describes systems and techniques for implementing a touchscreen. These systems and/or techniques enable processing of a signal generated from one or more sensors of a touchscreen to reduce noise and increase accuracy.