Abstract: An image recognition method includes: inputting a sample image data set into an original neural network model in advance; and for each convolution layer of the original neural network model, using a feature map of the sample image data set in the current layer as a reconstruction target, firstly obtaining an update weight of a convolution kernel by using a kernel set construction method, then calculating an input channel combination having a minimum reconstruction error, cutting a redundant input channel to obtain a compression result of the current convolution layer, and finally, splicing compression results of the convolution layers to generate an image recognition model. An image to be recognized is obtained and is inputted into the image recognition model, and an output result of the image recognition model is used as an image recognition result of the image.

Abstract: The provided is a visualized biological trawl net system based on a submersible. The visualized biological trawl net system based on a submersible includes a fixation frame, and further includes a collection net, a filtering mechanism, a fish school observation camera and a central control system. The collection net is arranged at a rear of the fixation frame and used for collecting marine creatures; the filtering mechanism is arranged inside the fixation frame, and the filtering mechanism is used for screening marine creatures entering the collection net; the fish school observation camera is arranged at a top of the fixation frame, and the fish school observation camera is used for observing marine creatures; and the central control system is electrically connected with the fish school observation camera and used for controlling the fish school observation camera to monitor marine creatures.

Abstract: An image processing method is provided. Since in the present invention, the first quantization threshold obtained by the saturated mapping method and the second quantization threshold obtained by the unsaturated mapping method are weighted, it is equivalent to fusing two quantization threshold values. The obtained optimal quantization threshold can be applied to most activations, to more effectively retain the effective information of the activations and use in subsequent image processing, thereby improving the accuracy of inference computations of the quantified deep neural network on low-bit-width hardware platforms. An image processing device and apparatus have the same beneficial effects as the above image processing method.

Abstract: A battery including a housing assembly, an electrode assembly, and a conductive assembly. The housing assembly includes a first housing fixed on a second housing. The second housing is provided with a cavity and includes a bottom plate. The electrode assembly is disposed in the cavity. The bottom plate contains an elastic portion protruding toward the first housing. The electrode assembly includes a first electrode plate and a second electrode plate stacked together. A surface of the first electrode plate facing back from a center of the electrode assembly contains a first blank foil region uncoated with a first active material and electrically connected to the conductive assembly. A surface of the second electrode plate facing back from the center of the electrode assembly contains a second blank foil region uncoated with a second active material. The elastic portion is electrically connected to the second blank foil region.

Abstract: A method for recognizing emotion, a training method, apparatuses, a device, a storage medium and a product. The method for recognizing emotion includes: acquiring to-be-recognized spiking sequences corresponding to video information; and recognizing the to-be-recognized spiking sequences by using a spiking neural network emotion recognition model, so as to obtain a corresponding emotion category.

Abstract: The image processing apparatus comprises at least a first arithmetic unit and a second arithmetic unit. The first arithmetic unit and the second arithmetic unit are cascaded by means of a serial transceiver, and are configured to execute the image processing method in parallel.

Abstract: A neural network compression method whereby forward inference is performed on target data by using a target parameter sharing network to obtain an output feature map of the last convolutional module, a channel related feature is extracted from the output feature map, the extracted channel related feature and a target constraint condition are input into a target meta-generative network, and an optimal network architecture under the target constraint condition is predicted by using the target meta-generative network to obtain a compressed neural network model.

Abstract: The disclosure describes methods of synthesis of pyridazinone compounds as thyroid hormone analogs and their prodrugs. Preferred methods according to the disclosure allow for large-scale preparation of pyridazinone compounds having high purity. In some embodiments, preferred methods according to the disclosure also allow for the preparation of pyridazinone compounds in better yield than previously used methods for preparing such compounds. Also disclosed are morphic forms of a pyridazinone compound. Further disclosed is a method for treating resistance to thyroid hormone in a subject having at least one TR? mutation.

Abstract: The disclosure describes methods of synthesis of pyridazinone compounds as thyroid hormone analogs and their prodrugs. Preferred methods according to the disclosure allow for large-scale preparation of pyridazinone compounds having high purity. In some embodiments, preferred methods according to the disclosure also allow for the preparation of pyridazinone compounds in better yield than previously used methods for preparing such compounds. Also disclosed are morphic forms of a pyridazinone compound. Further disclosed is a method for treating resistance to thyroid hormone in a subject having at least one TR? mutation.

Abstract: A deep-sea crawling robot and a crawling method thereof are provided. The deep-sea crawling robot includes a crawling robot body and a countertop, wherein a top of the crawling robot body is fixedly connected to the countertop, left and right sides of a bottom of the crawling robot body are respectively provided with crawler belts, and left and right sides of the countertop are respectively provided with weight-increasing auxiliary crawling mechanisms; and an outer plate is fixedly connected to a front side or back side of the countertop, a multifunctional propelling mechanism is arranged on the outer plate, and a rapid dewatering mechanism is arranged in a chamber of the countertop. The deep-sea crawling robot can not only accelerate the propulsion and crawling speed when it is on the seabed but also further enhance its stability during the crawling process on the seabed.

Abstract: A battery includes a conductive housing, an insulation layer, a pole, and a cell unit. The insulation layer is disposed in the housing, the pole penetrates through the insulation layer and is insulated from the housing. The electrode plates include a first and a second electrode plates. The first electrode plate includes a first current collector with a first active layer being disposed on one side of it facing the separator, an uncoated region on the other side is to form a first electrode output, which is electrically connected to the pole or the housing. The second electrode plate includes a second current collector, where a second active layer is disposed on at least one side of the second current collector, the second current collector forms a second electrode output end, which is electrically connected to the other one of the housing or the pole.

Abstract: A hardware environment-based data operation method, apparatus and device, and a storage medium. The method includes: determining data to be operated and target hardware, wherein the target hardware is a hardware resource that needs to perform convolution computation on the data to be operated currently; determining the maximum number of channels in which the target hardware executes parallel computation, and determining a data layout corresponding to the maximum number of channels to be an optimal data layout; and converting the data layout of the data to be operated into the optimal data layout, and performing the convolution computation on the data to be operated by using the target hardware after the conversion is completed.

Abstract: A method, an apparatus, and a device for target detection in consecutive images, and a computer-readable storage medium. A second frame is divided into multiple sub-images, before a target in the second frame in a video sequence is detected through a target-detecting network model. A first frame is searched, according to a preset rule for motion estimation, for a corresponding image block matched with each sub-image. Pixels of a sub-image, of which the matched image block is found in the first frame, are replaced with preset background pixels. Hence, a target repeating in both frames is replaced. Finally, the second frame subject to the replacement is inputted in to the target-detecting network model, to obtain a bounding box of a target object of the second frame and a category of such target object. An algorithm for target detection in consecutive images is optimized.

Abstract: An image recognition method and apparatus based on a systolic array, and a medium are disclosed. The method includes: converting obtained image feature information into a one-dimensional feature vector; converting an obtained weight matrix into a one-dimensional weight vector, and allocating a corresponding weight group to each node in a trained three-dimensional systolic array model; performing multiply-accumulate of the feature vector and a weight value on the one-dimensional feature vector in parallel by using the three-dimensional systolic array model, to obtain a feature value corresponding to each node, the feature values with different values reflecting article categories contained in an image; and determining an article category contained in the image according to the feature value corresponding to each node and a pre-established corresponding relationship between the feature value and the article category.

Abstract: Provided are a method, apparatus and device for extracting image features, and a storage medium. The method includes: obtaining parameters to be quantized of a network layer in a neural network model (S101); determining whether values of the parameters to be quantized are all positive numbers (S102); when the values of the parameters to be quantized are all positive numbers, executing, based on an asymmetric linear quantization logic, a quantization operation on the parameters to be quantized (S103); when the values of the parameters to be quantized are not all positive numbers, executing, based on a symmetric linear quantization logic, a quantization operation on the parameters to be quantized (S104); and extracting features of an input image by using the neural network model for which the quantization operation has been executed (S105).

Abstract: The present invention provides a method and apparatus for data caching. The method comprises: output matrixes are acquired one by one, a plurality of acquired output matrixes are written alternately into two queue sets of a first cache unit according to a sequence in which the output matrixes are acquired, and the output matrixes stored line by line in a first cache unit are written into a second cache unit one by one, according to the sequence in which the output matrixes are written into the second cache unit, valid data of each output matrix of the second cache unit is determined one by one according to preset parameters, and the valid data of each output matrix is written into a third cache unit, and the valid data of the output matrixes stored in the third cache unit are configured to be sequentially written into a memory according to a sequence in which the valid data are written into the third cache unit.

Abstract: An image recognition method and device, and a computer-readable storage medium. The method includes: inputting a sample image data set into an original neural network model in advance; and for each convolution layer of the original neural network model, using a feature map of the sample image data set in the current layer as a reconstruction target, firstly obtaining an update weight of a convolution kernel by using a kernel set construction method, then calculating an input channel combination having a minimum reconstruction error, cutting a redundant input channel to obtain a compression result of the current convolution layer, and finally, splicing compression results of the convolution layers to generate an image recognition model.

Abstract: An image recognition method and apparatus based on a systolic array, and a medium are disclosed. The method includes: converting obtained image feature information into a one-dimensional feature vector; converting an obtained weight matrix into a one-dimensional weight vector, and allocating a corresponding weight group to each node in a trained three-dimensional systolic array model; performing multiply-accumulate of the feature vector and a weight value on the one-dimensional feature vector in parallel by using the three-dimensional systolic array model, to obtain a feature value corresponding to each node, the feature values with different values reflecting article categories contained in an image; and determining an article category contained in the image according to the feature value corresponding to each node and a pre-established corresponding relationship between the feature value and the article category.

Abstract: A system for accelerating an RNN network including: a first cache, which is used for outputting Wx1 to WxN or Wh1 to WhN in parallel in N paths in a cyclic switching manner, and the degree of parallelism is k; a second cache, which is used for outputting xt or ht-1 in the cyclic switching manner; a vector multiplication circuit, which is used for, by using N groups of multiplication arrays, respectively calculating Wx1xt to WxNxt, or respectively calculating Wh1ht-1 to WhNht-1; an addition circuit, which is used for calculating Wx1xt+Wh1ht-1+b1 to WxNxt+WhNht-1+bN; an activation circuit, which is used for performing an activation operation according to an output of the addition circuit; a state updating circuit, which is used for acquiring ct-1, calculating ct and ht, updating ct-1, and sending ht to the second cache; a bias data cache; a vector cache; and a cell state cache.

Abstract: A neural network compression method, apparatus and device, and a storage medium are provided. The method includes: performing forward inference on target data by using a target parameter sharing network to obtain an output feature map of the last convolutional module; extracting a channel related feature from the output feature map; inputting the extracted channel related feature and a target constraint condition into a target meta-generative network; and predicting an optimal network architecture under the target constraint condition by using the target meta-generative network to obtain a compressed neural network model. By using the technical solution, the computation load of a performance evaluation process of a neural architecture search may be reduced, and the speed of the searching for a high-performance neural network architecture may be increased.