5results about How to "Resolve confusion" patented technology

Embodiments of the present specification provide a method and apparatus for training a generation model and generating training samples for a text classifier. In the method for training the generation model, first processing and second processing are performed on a first text sample. The first processing includes determining a semantic vector of the first text sample by a first encoder. A first category of the first text sample is predicted based on the semantic vector by a text classifier, and a first prompt text corresponding to the first category is constructed. The second processing includes determining a first discrete vector corresponding to the first text sample in a target vector space by a second encoder. A reconstructed text of the first text sample is determined based on the first prompt text and the first discrete vector by a decoder. The generation model is trained based on a reconstruction loss determined based on the first text sample and the reconstructed text.

This application relates to the field of fire monitoring technology and provides a multimodal remote sensing fire smoke recognition method. First, remote sensing images are input into a target detection model, and features are extracted through multiple branches to output directional detection information. These multiple branches include a first branch and a second branch. The first branch extracts features from the input visible light spectral image, near-infrared image, and short-wave infrared image respectively; the second branch extracts features from the input thermal infrared image across multiple bands respectively. The directional detection information includes the category label, geometric center point coordinates, and rotated bounding box of the fire smoke target. Then, the directional detection information is input into a SAM model to segment the input remote sensing image, outputting a segmentation mask of the fire smoke target. Finally, the attribute information of the fire smoke target region is determined based on the input segmentation mask, including the coverage area and shape attributes of the fire smoke target.

This invention discloses a method and system for snow and ice identification based on remote sensing images. The method includes: acquiring remote sensing images of the target area and performing sensor-specific preprocessing and topographic radiometric correction to obtain a standardized surface reflectance image; extracting improved spectral index features, scale-adaptive texture features, and topographic occlusion compensation features based on multi-level feature coupling rules to construct a multi-dimensional feature set; generating a snow and ice probability map through a feature pyramid network based on an attention mechanism; generating an initial snow and ice mask using a dynamic window adaptive threshold segmentation strategy, combined with elevation zonation constraints and multi-temporal change trajectories; and finally, removing transient coverage and noise through spatiotemporal consistency joint optimization to generate an accurate snow and ice coverage map. This invention effectively solves the problem of spectral confusion between snow and ice, clouds, and bare rock, overcomes the influence of topographic shadows and transient interference, and significantly improves the accuracy and reliability of snow and ice identification in complex environments.

The invention relates to the technical field of natural language processing, in particular to a dialogue abstract generation method based on a large-scale language model LLM auxiliary tag and two-dimensional centrality and a storage medium thereof.The method comprises the steps that the large-scale language model LLM is used for generating an intention tag of a dialogue utterance, and a training set TS with a high-quality tag is constructed after confidence coefficient screening; taking a pre-training model as a backbone, and training a deep learning network model fusing a large-scale language model LLM auxiliary label and two-dimensional centrality based on the training set TS; inputting a to-be-processed dialogue into the trained deep learning network model F, and outputting a dialogue abstract; according to the method, the accuracy, the integrity and the interpretability of the dialogue abstract can be improved.

The utility model belongs to the technical field of plastic product production unloading, especially a kind of unloading device for plastic product production, including unloading guide frame, the front side fixed mounting of unloading guide frame has fixed frame, cooling setting box is cooperated with fan, cooling cooling in the transmission process of plastic product, accelerate product setting, avoid the deformation problem caused by temperature too high;Compared with the function that only transmission unloading of traditional unloading device can be carried out, the design ensures the form stability of plastic product in the unloading stage, improves product quality stability;Industrial high-definition camera and fill light can carry out quality detection to plastic product, realize automatic classification in combination with subsequent transmission structure: qualified product is transferred to second conveying belt by suction cup after being transmitted by first conveying belt;Unqualified product falls into collecting frame through unqualified product guide frame;This automatic classification replaces manual screening, solves the problem that qualified product and unqualified product are confused, reduces the artificial cost of warehouse link.