46results about How to "Reduce model size" patented technology

The disclosure discloses a neural-network computing system. The system includes: an I / O interface, which is used for I / O of data; a memory, which is used for temporarily storing a multi-layer artificial-neural-network model and neuron data; an artificial-neural-network chip, which is used for executing multi-layer artificial-neural-network operation and a back-propagation training algorithm thereof, wherein data and a program from a central processing unit (CPU) are accepted, and the above-mentioned multi-layer artificial-neural-network operation and the back-propagation training algorithm thereof are executed; the central processing unit CPU, which is used for data transportation and starting / stopping control of the artificial-neural-network chip, is used as an interface of the artificial-neural-network chip and external control, and receives results after execution of the artificial-neural-network chip. The disclosure also discloses a method of applying the above-mentioned system forartificial-neural-network compression encoding. According to the system, a model size of an artificial neural network can be effectively reduced, data processing speed of the artificial neural network can be increased, power consumption can be effectively reduced, and a resource utilization rate can be increased.

A remote control toy system includes a plurality of sets. Each set includes a controller, and a model controlled based on data transmitted from the controller so as to correspond to an operation of a user. An accessory device is provided as a device independent from the controllers and models. The accessory device can conduct data communication with the controllers and models. Each of the controllers, the models, and the accessory device includes a radio communication module serving as a device for executing the data communication and can conduct bilateral data communication based on the Bluetooth standards, and a control device for implementing various controls based on data communication conducted through the radio communication module.

An automatic portrait division method disclosed by the present invention comprises the step model training, the portrait initial division and the portrait accurate division, wherein the model training comprises the steps of obtaining the training data, inputting the training data in a neural network, and learning to obtain a network parameter mode; after the neural network converges, storing the network parameter model to finish the model training; the portrait initial division comprises the steps of loading the network parameter model in a mobile client; utilizing the mobile client to obtain an input image; inputting the input image in the network parameter model to obtain a portrait division initial image; and the portrait accurate division comprises the steps of removing the mistakenly divided isolated areas for the portrait division initial image, and retaining a communicated area; utilizing an image matting algorithm to divide the details of the communicated area, and highlighting the edges to obtain a portrait division final image. The automatic portrait division method of the present invention can divide the portrait automatically and accurately, is short in time consumption and small in occupied space, and is suitable for the mobile client.

A pitching arrangement for a model helicopter, comprising a securing base, a plurality of servers supported by the securing base, a main connecting shaft communicated to the servers, and has a supporting ring, a control device operatively provided between the servers, and is axially mounted to the main connecting shaft, and a plurality of connecting rods connecting the servers and the control device through the main connecting shaft, wherein the supporting ring is mounted on the main connecting shaft, wherein the connecting rods are connected to the supporting ring of the main connecting shaft.

The invention relates to a sectional navigation lane changing method and system, computer equipment and a storage medium, and the method comprises the steps: employing an LSTM network to judge whether an adjacent target lane meets a lane changing condition or not according to the speed of a vehicle at the current moment, the speed difference and distance between the vehicle and surrounding vehicles, and other information, and if not, continuing to collect related information and inputting the related information into the LSTM network; if so, acquiring a lane center line of the adjacent target lane, selecting a plurality of points on the lane center line, and acquiring position information of the plurality of points; acquiring distance information between the vehicle and the lane center line; inputting the position information of the plurality of points and the distance information between the vehicle and the lane center line into a CNN network for convolution calculation to obtain a target steering wheel angle; and finally, sending the target steering wheel angle to an automatic driving control unit of the vehicle to drive the automatic driving control unit to control the vehicle to change lanes according to the steering wheel angle. According to the invention, the lane changing process is more intelligent and accords with human driving habits.