This document describes techniques and systems for a road-perception system to estimate a belief and plausibility of lanes in a road model. The road-perception system models lanes of a road using basis bands to represent lane sections. The basis bands comprise points representing a polyline with a width at each point. The use of basis bands results in lower-computational requirements and a more stable road model than grid-based representations of a driving environment. The road-perception system also determines a belief mass associated with the lane sections. The road-perception system then computes, using the belief masses, a belief parameter and a plausibility parameter associated with proposed lanes of the road model. In this way, the described techniques and systems can provide an accurate and reliable road model and quantify uncertainty therein.

Computer automated interactive activity recognition based on keypoint detection includes retrieving, by one or more processors, a temporal sequence of image frames from a video recording. The one or more processors identify first and second keypoints in each of the image frames in the temporal sequence using machine learning techniques. The first keypoints are associated with an object in the temporal sequence of image frames while the second keypoints are associated with an individual interacting with the object. The one or more processors combine the first keypoints with the second keypoints and extract spatial-temporal features from the combination that are used to train a classification model based on which interactive activities can be recognized.

Feature descriptor matching is reformulated into a graph-matching problem. Keypoints from a query image and a reference image are initially matched and filtered based on the match. For a given keypoint, a feature graph is constructed based on neighboring keypoints surrounding the given keypoint. The feature graph is compared to a corresponding feature graph of a reference image for the matched keypoint. Relocalization data is obtained based on the comparison.

An activity recognition system is disclosed. A plurality of temporal features is generated from a digital representation of an observed activity using a feature detection algorithm. An observed activity graph comprising one or more clusters of temporal features generated from the digital representation is established, wherein each one of the one or more clusters of temporal features defines a node of the observed activity graph. At least one contextually relevant scoring technique is selected from similarity scoring techniques for known activity graphs, the at least one contextually relevant scoring technique being associated with activity ingestion metadata that satisfies device context criteria defined based on device contextual attributes of the digital representation, and a similarity activity score is calculated for the observed activity graph as a function of the at least one contextually relevant scoring technique, the similarity activity score being relative to at least one known activity graph.

An activity recognition system is disclosed. A plurality of temporal features is generated from a digital representation of an observed activity using a feature detection algorithm. An observed activity graph comprising one or more clusters of temporal features generated from the digital representation is established, wherein each one of the one or more clusters of temporal features defines a node of the observed activity graph. At least one contextually relevant scoring technique is selected from similarity scoring techniques for known activity graphs, the at least one contextually relevant scoring technique being associated with activity ingestion metadata that satisfies device context criteria defined based on device contextual attributes of the digital representation, and a similarity activity score is calculated for the observed activity graph as a function of the at least one contextually relevant scoring technique, the similarity activity score being relative to at least one known activity graph.

A method for processing a candidate digital image includes defining a set of noteworthy points in the candidate digital image. A set of at least three noteworthy points is selected to comprise a notable departure point, a notable arrival point, and a third notable point not aligned with the notable departure point and the notable arrival point. A set of at least one route, between the notable departure point and the notable arrival point, is defined. The route passes through all of the selected notable points. Local characteristics, of the pixels located along the route, are extracted. The signal, corresponding to the variation in the magnitude of the local characteristics as a function of each pixel along each defined route, is recorded in the form of a fingerprint.

Technology is disclosed herein for enhanced similarity search. In an implementation, a search environment includes one or more computing hardware, software, and/or firmware components in support of enhanced similarity search. The one or more components identify a modality for a similarity search with respect to a query object. The components generate an embedding for the query object based on the modality and based on connections between the query object and neighboring nodes in a graph. The embedding for the query object provides the basis for the search for similar objects

The present disclosure discloses a method and apparatus for generating a human body three-dimensional model, a device and a storage medium. The method may include: receiving a single human body image, and extracting an SMPL human body three-dimensional model corresponding to the human body image and a PIFu human body three-dimensional model corresponding to the human body image; matching the SMPL human body three-dimensional model with the PIFu human body three-dimensional model to obtain a matching result; determining a vertex of the SMPL human body three-dimensional model closest to a vertex of the PIFu human body three-dimensional model based on the matching result to obtain a binding weight of the vertex of the PIFu human body three-dimensional model and each skeleton point of the SMPL human body three-dimensional model; and outputting a drivable human body three-dimensional model.

System and methods are provided for predicting a roadway feature. A geometric pattern in a roadway network is selected. The geometric pattern comprises one or more first links and one or more first nodes. A similar pattern to the geometric pattern is identified that comprises one or more second links and one or more second nodes in the roadway network. One or more features for the geometric pattern are identified. An absence of a feature in the similar pattern is determined based on the one or more features for the geometric pattern. A notification relating to the absence of the feature is generated.

In some implementations, there is provided a method, which includes transforming, by the recommendation system, a first data set into the principal component analysis domain; rotating, by the recommendation system, the transformed first data first data set into a common axis system; comparing, by the recommendation system, the rotated, transformed first data set to at least one of a plurality of reference data sets having been rotated into the common axis system; and identifying, by the recommendation system, at least one reference data set, the identifying based on the comparing. Related systems, methods, and articles of manufacture are also disclosed.