75results about How to "Avoid duplicate fetches" patented technology

The invention discloses a pedestrian re-identification method based on multi-channel attention features, comprising the following steps: 1) constructing a convolutional neural network model based on channel attention, and pre-training a backbone network; 2) extracting pedestrian pictures in the backbone network 3) The weighted vector is multiplied by the output feature of the backbone network, and then added with it to obtain the channel attention feature; 4) Multiple attention features are repeatedly extracted, and the sea The Ringer distance is used to regularize the feature diversity; 5) The attention feature is input into the fully connected layer and the classifier, and the training minimizes the cross entropy loss and metric loss; 6) The test set image is input into the trained model to extract features, and the metric Sorting achieves pedestrian re-identification. The invention extracts the discriminative features of pedestrians based on the attention mechanism, limits the repeated extraction of similar attention features, and effectively improves the accuracy and robustness of pedestrian re-identification.

The embodiment of the invention discloses a method for integrated detection and fine-grained identification of a remote sensing image ship target. The method includes: acquiring a training sample set; determining a feature extraction shared model integrated in a remote sensing image processing model, a location detection branch model for outputting the location information of the target object, and a type recognition branch model for outputting the type information of the target object ; Adopt the shared image features extracted from the training samples by the feature extraction shared model to alternately train the position detection branch model and the type recognition branch model, so as to complete the training of the remote sensing image processing model when the preset convergence conditions are met. . By adopting the solution of the present application, it is possible to avoid repeated extraction of the shared part of the image features during training or use of the position detection branch model and the type recognition branch model, avoid additionally increasing the task load of this part of the extraction operation, and not only greatly reduce the model training. amount of tasks.

The invention discloses a message communication method and a system. The message communication method includes the following steps. Firstly, before a sending client terminal sends information comprising images, the fact that whether an image managing server stores the images or not is judged. If negative, the images are uploaded to the image managing server and provided corresponding identification. The image managing server stores the images with identification of the images as a search main key. Secondly, the sending client terminal sends the message comprising the images to a receiving client terminal. The images in the sent message are expressed as descriptive information of the images. The descriptive Information comprises identification of the images. Thirdly, the images are obtained based on the identification of the images. The receiving client terminal displays the images at a corresponding position of the message. The message communication method and the system can use any images in message communication and save network resources.

The invention discloses a measurement parameter hybrid framing method of a missile-borne recorder. According to the method, measurement parameters are received by use of an FPGA chip, wherein digital measurement parameters directly go into the FPGA chip through a serial port for storage in FIFO2; and analog measurement parameters go into the FPGA chip through a multiplexer and an A / D converter which are sequentially connected and are finally written into an externally connected FLASH memory through the FPGA. The analog measurement parameters are acquired by use of an A / D converter with n channels, after fully packed data packets are obtained through acquisition, the data packets are sent to the FPGA; the FPGA, according to an acquisition frequency, extracts data from the fully packed data packets and carries out framing, and data after the framing is stored in FIFO1; and then, in a GPIO interruption service program, the data of the FIFO1 and the FIFO2 are respectively written into different pages of the externally connected FLASH memory. The method provided by the invention simplifies the frame format, improves the effective signal transmission bandwidth, reduces the complexity of a multipath switch control algorithm and data processing software, and has certain flexibility.