Microsoft Technology Licensing, LLC

Norberto Goussies

Pablo Luis Sala

Zicheng Liu

A system includes a memory and a processor configured to select a set of scene point pairs, to determine a respective feature vector for each scene point pair, to find, for each feature vector, a respective plurality of nearest neighbor point pairs in feature vector data of a number of models, to compute, for each nearest neighbor point pair, a respective aligning transformation from the respective scene point pair to the nearest neighbor point pair, thereby defining a respective model-transformation combination for each nearest neighbor point pair, each model-transformation combination specifying the respective aligning transformation and the respective model with which the nearest neighbor point pair is associated, to increment, with each binning of a respective one of the model-transformation combinations, a respective bin counter, and to select one of the model-transformation combinations in accordance with the bin counters to detect an object and estimate a pose of the object.

Real-time, model-based object detection and pose estimation

Described is a technology by which video, which may be relatively high-resolution video, is efficiently processed to determine whether the video contains a specified action. The video corresponds to a spatial-temporal volume. The volume is searched with a top-k search that finds a plurality of the most likely sub-volumes simultaneously in a single search round. The score volumes of larger spatial resolution videos may be down-sampled into lower-resolution score volumes prior to searching.

Growing community of inventors

Buenos Aires, Argentina

Norberto Goussies