6 results about "Natural language processing" patented technology

The invention relates to an indoor scene semantic annotation method based on RGB-D data. According to the method, a coarse-to-fine global recursion feedback semantic annotation framework based on the RGB-D data is built, in addition, the whole semantic annotation framework is divided into two major parts including the coarse-granularity region stage semantic label deduction and fine-granularity pixel stage semantic label refinement. The framework is different from the traditional region stage or pixel stage semantic annotation framework, the framework rebuilds the relationship between the coarse-granularity region stage semantic label deduction and the fine-granularity pixel stage semantic annotation, and a reasonable global recursion feedback mechanism is introduced, so that the coarse-granularity region stage semantic annotation result and the fine-granularity pixel level semantic annotation result realize the alternate iterative updating optimization. Through adopting the mode, the multi-mode information of different region layers in the scene images is better merged, and the general problem that an annotation base element is difficult to be properly selected in the traditional indoor scene semantic annotation scheme is solved to a certain degree.

The invention discloses a receipt automatic identification method based on OCR. The method comprises the steps of text extraction through OCR, document type recognition, document template recognition,universal recognition rule recognition of character meanings and recognition of character collection in sequence, completing text recognition of receipts in logistics transportation, and meanwhile improving recognition precision through error correction in the license plate ID recognition process. According to the method, the effect of extracting the cargo transportation elements from a section of disordered and meaningless characters is achieved, so that the text recognition accuracy and the logistics transportation detection working efficiency are improved.

The invention discloses a model training and speech synthesis method and device, equipment and a medium, and relates to the technical field of artificial intelligence, deep learning and speech. The specific implementation scheme is that the model training method comprises the steps: obtaining a sample text in a training data set; determining label information corresponding to the sample text basedon an acoustic model trained by adopting an unsupervised training method in advance, wherein the label information comprises style information and/or role information; training a text classificationmodel based on the sample text and label information corresponding to the sample text, wherein the text classification model is used for outputting corresponding label information according to an input text. According to the embodiment of the invention, the technical effect of automatically determining the label information corresponding to the sample text is realized, the label labeling accuracyand efficiency are improved, and the training speed of the text training model is correspondingly improved.

The invention provides a text data desensitization method. The text data desensitization method comprises the steps of obtaining text data, processing the text data through a preset dictionary tree and/or a preset regular expression to obtain a first sensitive entity word of the text data, processing the text data through a preset model to obtain a label of each character in the text data, determining a second sensitive entity word of the text data according to the label of each character in the text data, determining sensitive entity words of the text data according to the first sensitive entity words of the text data and the second sensitive entity words of the text data, and carrying out desensitization processing on the sensitive entity words of the text data. In this way, it is determined that the sensitive entity words of the text data are more accurate by utilizing mutual complementation and proofreading of the first sensitive entity words and the second sensitive entity words,so that the result of desensitization processing on the text data is more accurate, desensitization processing on non-sensitive entity words is avoided, and the user experience is improved.

Embodiments of the invention provide a dictionary format generation method and an electronic device, which are used for processing special characters in a text and improving the accuracy of segmentinga field containing a special character string in text analysis. The method includes the steps of obtaining a plurality of textual information pieces from at least one data source, each textual information piece of the plurality of textual information pieces including a special character, and the special characters including numbers and/or letters; extracting at least one semantic segment relatedto the special characters in the plurality of textual information pieces, wherein each semantic segment in the at least one semantic segment includes the special characters and associated characters adjacent to the special characters and the number of characters of the associated characters is less than or equal to a preset number; and determining at least one dictionary format according to the atleast one semantic segment, the at least one dictionary format being used for representing distribution rules of the special characters in the corresponding semantic segment.

The invention provides an ancient book proper name recognition method and device. According to the method, an ancient book text is acquired, word segmentation processing is performed on the ancient book text based on an ancient book statistical model word segmentation algorithm, and multiple proper names are obtained; each proper name is classified based on a pre-saved proper name knowledge base;for each proper name, whether the proper name is a book name in the pre-saved proper name knowledge base is judged; and if not, for each proper name, when it is judged that the proper name only belongs to one of classification categories, the proper name is saved as a target proper name of the ancient book text. Through the method, the problems that traditional manual proper name annotation consumes time and labor and traditional machine annotation has a poor effect can be relieved, the precision and quality of ancient book proper name recognition are improved, and therefore the actual application demand in ancient book sorting and publishing work is met.