3 results about "Point correspondence" patented technology

In an example method of training a neural network for performing visual odometry, the neural network receives a plurality of images of an environment, determines, for each image, a respective set of interest points and a respective descriptor, and determines a correspondence between the plurality of images. Determining the correspondence includes determining one or point correspondences between the sets of interest points, and determining a set of candidate interest points based on the one or more point correspondences, each candidate interest point indicating a respective feature in the environment in three-dimensional space). The neural network determines, for each candidate interest point, a respective stability metric and a respective stability metric. The neural network is modified based on the one or more candidate interest points.

The method comprises the following steps: obtaining a set of points based on historical information corresponding to a target object; extracting a plurality of sets of sentences from at least one explanation text corresponding to the target object based on the plurality of points in the set of points; determining a plurality of weight values corresponding to the plurality of points based on a user portrait corresponding to a target user, and determining the order of the plurality of points based on the weight values; sorting the sets of sentences corresponding to the plurality of points of the target object in the order to obtain a description text corresponding to the target user of the target object; and obtaining a long text through the extracted sets of sentences, and determining the order of the sets of sentences based on the weight values determined based on the user portrait, thereby enhancing the interaction with the user and realizing personalized generation of the corresponding description text for the target user.

This invention discloses a point-of-interest recommendation method, apparatus, electronic device, and storage medium. The method includes: acquiring a user's current captured image information; comparing the user's captured image with multiple file information in a reference file set to determine the target file information corresponding to the user's captured image; determining a corresponding trip frequency matrix based on the target file information and the current point-of-interest information; the target file information includes captured image information of the user at multiple trip points, and the trip frequency matrix describes the number of times the user appears at each trip point at different times, with pre-calculated distance similarity and time similarity between multiple trip points; determining target interest points based on the trip frequency matrix, and recommending points of interest based on the target interest points. This makes the point-of-interest recommendation significantly more reasonable, capable of addressing user interest needs at different times and locations, and improving the accuracy and recall of the recommendation.