Abstract: A performing device of an impulse-like gesture recognition system executes an impulse-like gesture recognition method. A performing procedure of the impulse-like gesture recognition method includes steps of: receiving a sensing signal from a sensing unit; determining a prediction with at least one impulse-like label according to the sensing frames by a deep learning-based model; and classifying at least one gesture event according to the prediction. The gesture event is classified to determine the motion of the user. Since the at least one impulse-like label is used to label at least one detection score of the deep learning-based model, the detection score is non-decreasing, reaction time of the at least one gesture event for an incoming gesture is fast, rapid consecutive gestures are easily decomposed, and an expensive post-processing is not needed.

Abstract: A method of customizing a hand gesture provides a touch screen, a computing unit connected with the touch screen, and a hand gesture database connected with the computing unit, and the method includes the following steps: recording a hand gesture trajectory data input of an input hand gesture on the touch screen; converting the hand gesture trajectory data input into a 2D trajectory graph of the input hand gesture; the computing unit sequentially reads a 2D hand gesture reference graph from the hand gesture database, and correspondingly generates a 2D hand gesture reference graph set for the 2D hand gesture reference graph read, and then the computing unit compares the similarity between the 2D trajectory graph of the input hand gesture and each reference graph in the 2D hand gesture reference graph set to determine whether the input hand gesture is already in the hand gesture database.

Abstract: A generic gesture detecting method executed by a generic gesture detecting device includes steps of: receiving a current sensing signal from a sensing unit; generating a current image according to the current sensing signal; determining whether the current image is similar with a stored image stored in a memory unit; when the current image is similar with the stored image, detecting the current image and the stored image to be a gesture signal; when the current image is different from the stored image, storing the current image into the memory unit, and returning to the step of receiving a current sensing signal. Since the generic gesture detecting device can automatically detect the gesture signal, the user may not need to enable a detecting time period before implementing a command motion. Therefore, the user can make the command motion without enabling the detecting time period, and the convenience can be increased.

Abstract: An earphone device having gesture recognition functions includes a gesture recognition element, a signal transmission unit, and a voice output element. The gesture recognition element includes a transmission unit, a reception chain and a processing unit. The transmission unit transmits a transmission signal to detect the gesture. The reception chain receives a gesture signal to generate a feature map data. The processing unit is coupled to the reception chain for receiving the feature map data and utilizes an identification algorithm to recognize gesture according to the feature map data to generate a gesture controlling signal. The signal transmission unit receives and transmits the gesture controlling signal to an electronic device. The processing unit receives a controlling action generated by the electronic device according to the gesture controlling signal via the signal transmission unit.

Abstract: A range Doppler angle detection method executed by a range Doppler angle detection device includes steps of: receiving a first sensing signal and a second sensing signal; performing 1D Fast Fourier Transform (FFT) and 2D FFT to the first sensing signal for calculating one first 2D FFT map; performing the 1D FFT and the 2D FFT to the second sensing signal for calculating one second 2D FFT map; picking up one column of the first 2D FFT map and one column of the second 2D FFT map according to a given Doppler index; performing the 3D FFT to the picked column of the first 2D FFT map and the picked column of the second 2D FFT map for calculating a range Doppler angle. Therefore, a computation loading of the gesture recognition function can be reduced.

Abstract: A generic gesture detecting method executed by a generic gesture detecting device includes steps of: receiving a current sensing signal from a sensing unit; generating a current image according to the current sensing signal; determining whether the current image is similar with a stored image stored in a memory unit; when the current image is similar with the stored image, detecting the current image and the stored image to be a gesture signal; when the current image is different from the stored image, storing the current image into the memory unit, and returning to the step of receiving a current sensing signal. Since the generic gesture detecting device can automatically detect the gesture signal, the user may not need to enable a detecting time period before implementing a command motion. Therefore, the user can make the command motion without enabling the detecting time period, and the convenience can be increased.

Abstract: A range Doppler angle detection method executed by a range Doppler angle detection device includes steps of: receiving a first sensing signal and a second sensing signal; performing 1D Fast Fourier Transform (FFT) and 2D FFT to the first sensing signal for calculating one first 2D FFT map; performing the 1D FFT and the 2D FFT to the second sensing signal for calculating one second 2D FFT map; picking up one column of the first 2D FFT map and one column of the second 2D FFT map according to a given Doppler index; performing the 3D FFT to the picked column of the first 2D FFT map and the picked column of the second 2D FFT map for calculating a range Doppler angle. Therefore, a computation loading of the gesture recognition function can be reduced.

Abstract: The invention discloses an oscillator, including a voltage switching circuit, a voltage adjustment circuit and a frequency generation circuit. The voltage switching circuit receives an output voltage signal whereby the output voltage signal switches a first input voltage signal to a first voltage level signal and switches a second input voltage signal to a second voltage level signal. The voltage adjustment circuit receives the first voltage level signal and the second voltage level signal, whereby the first voltage level signal and the second voltage level signal generate the first adjustment voltage signal and the second adjustment voltage signal. The frequency generation circuit is connected to the voltage adjustment circuit, and receives the first adjustment voltage signal and the second adjustment voltage signal to generate the first output frequency signal and the second output frequency signal according to the first adjustment voltage signal and the second adjustment voltage signal.

Abstract: A performing device of an impulse-like gesture recognition system executes an impulse-like gesture recognition method. A performing procedure of the impulse-like gesture recognition method includes steps of: receiving a sensing signal from a sensing unit; determining a prediction with at least one impulse-like label according to the sensing frames by a deep learning-based model; and classifying at least one gesture event according to the prediction. The gesture event is classified to determine the motion of the user. Since the at least one impulse-like label is used to label at least one detection score of the deep learning-based model, the detection score is non-decreasing, reaction time of the at least one gesture event for an incoming gesture is fast, rapid consecutive gestures are easily decomposed, and an expensive post-processing is not needed.

Abstract: A method for compressing an initial weight matrix includes generating a first weight matrix and a second weight matrix according to the initial weight matrix where the initial weight matrix is a Kronecker product of a transposed matrix of the second weight matrix and the first weight matrix; optimizing the first and second weight matrixes to generate an optimized first weight matrix and an optimized second weight matrix; generating a processed data matrix according to an initial data matrix where the initial data matrix is vectorization of the processed data matrix; multiplying the processed data matrix by the optimized first weight matrix to generate a first product; multiplying the optimized second weight matrix by the first product to generate a second product; and vectorizing the second product. The initial weight matrix requires a larger memory space than a combined memory space of the first and second weight matrixes.

Abstract: A method for compressing an initial weight matrix includes generating a first weight matrix and a second weight matrix according to the initial weight matrix where the initial weight matrix is a Kronecker product of a transposed matrix of the second weight matrix and the first weight matrix; optimizing the first and second weight matrixes to generate an optimized first weight matrix and an optimized second weight matrix; generating a processed data matrix according to an initial data matrix where the initial data matrix is vectorization of the processed data matrix; multiplying the processed data matrix by the optimized first weight matrix to generate a first product; multiplying the optimized second weight matrix by the first product to generate a second product; and vectorizing the second product. The initial weight matrix requires a larger memory space than a combined memory space of the first and second weight matrixes.

Abstract: A gesture recognition system includes a Frequency modulated continuous waveform radar system. First and second channels of the signal reflected by the object are preprocessed and respectively sent to first and second feature map generators. A machine-learning accelerator is configured to receive output from the first and second feature map generators and form frames fed to a deep neural network realized with a hardware processor array for gesture recognition. A memory stores a compressed set of weights as fixed-point, low rank matrices that are directly treated as weights of the deep neural network during inference.

Abstract: A performing device of a gesture recognition system executes a performing procedure of a gesture recognition method. The performing procedure includes steps of: receiving a sensing signal; selecting one of sensing frames of the sensing signal; determining a soft label of the selected sensing frame; classifying a gesture event when the soft label of the selected sensing frame is approved. The gesture event is classified to determine the motion of the user. Therefore, the gesture recognition system does not need a predetermined time period to recognize the motion of the user. The time period for recognizing the motion of the user can be dynamical. A total time period for classifying a plurality of motions can be decreased, and the performance of the gesture recognition system can be improved.