Carnegie Mellon University

Phoenix Intangibles Holding Company

Jonathan Cagan

Philip LeDuc

Jay P McCormack

Daniel Clymer

Megan Rene Langdon

Francois Humbert

Su Yin

Janet Catov

Mark Whiting

Shraddha Premjibhai Joshi

Nayoung Kim

Koo Ho Shin

Robin Rutenbar

Leah Hoxie

Liron Pantanowitz

Stephen Van Marter Krotseng

Phil LeDuc

Eric Kemner

Simon Szykman

Daragh Joseph Byrne

Andrew Concilio

Ashish Kolli

A hierarchical deep-learning object detection framework provides a method for identifying objects of interest in high-resolution, high pixel count images, wherein the objects of interest comprise a relatively a small pixel count when compared to the overall image. The method uses first deep-learning model to analyze the high pixel count images, in whole or as a patchwork, at a lower resolution to identify objects, and a second deep-learning model to analyze the objects at a higher resolution to classify the objects.

Method for object detection using hierarchical deep learning

A data processing system is configured to pre-process data for a machine learning classifier. The data processing system includes an input port that receives one or more data items, an extraction engine that extracts a plurality of data signatures and structure data, a logical rule set generation engine configured to generate a data structure, select a particular data signature of the data structure, identify each instance of the particular data signature in the data structure, segment the data structure around instances of the particular data signature, identify one or more sequences of data signatures connected to the particular data signature, and generate a logical ruleset. A classification engine executes one or more classifiers against the logical ruleset to classify the one or more data items received by the input port.

Searching of data structures in pre-processing data for a machine learning classifier

Provided is a storage container including a base and a cartridge container attachable to the base. The cartridge container defines a plurality of internal chambers and includes a plurality of individually operable doors, each of the plurality of doors corresponding to one of the internal chambers. The storage container further includes a plurality of individually addressable actuators disposed in at least one of the base and the cartridge container, each of the plurality of actuators corresponding to one of the doors. A microprocessor is programmed or configured to receive at least one instruction from a server computer, and transmit a command, based on the at least one instruction, to actuate one of the plurality of actuators. Actuating one of the plurality of actuators causes the corresponding door to move from a closed position to an opened position. Systems and methods are provided for remote and immediate dispensing of articles.

Internet-connected storage container and system and method of dispensing articles

The invention generally relates to a shopping cart that could have: (i) a slidably movable lower container; (ii) a telescoping container; (iii) dual handles; (iv) a stand having a number of holders and a writing surface formed therein; (v) a wheel having a brake coupled thereto; (vi) a generally A-shaped frame; and/or (vii) a retractable child seat assembly.

Shopping cart

Parametric shape recognition is achieved through a decomposition of shapes into a hierarchy of subshapes ordered by their decreasing restrictions. Instances of each of the subshapes are individually located in the design shape and then reconstructed to form an instance of the entire shape. The basis for the hierarchy of subshapes can be specified by the designer or based on the default parameter relations that come from architectural and engineering knowledge. The levels of the hierarchy are defined so that the most constrained lines of a shape are those lines that the designer intended exactly. These most constrained lines have specified parametric relations to other line segments and those relations, if altered, will compromise the designer's intentions. Conversely, the lowest level of the hierarchy, which contains the least constrained line segments, only implies a specific connectivity between line segments, necessitating a vaster search. The parametric recognition of curved line shapes uses a two-step approach that first performs shape matching with an equivalent straight-line shape then checks those transformations for matching with the actual curved lines. This approach has advantages over just matching characteristic polygons in that it can match equivalent curves with differing characteristic polygons as well as emergent shapes.

Shape grammar interpreter

Parametric shape recognition is achieved through a decomposition of shapes into a hierarchy of subshapes ordered by their decreasing restrictions. Instances of each of the subshapes are individually located in the design shape and then reconstructed to form an instance of the entire shape. The basis for the hierarchy of subshapes can be specified by the designer or based on the default parameter relations that come from architectural and engineering knowledge. The levels of the hierarchy are defined so that the most constrained lines of a shape are those lines that the designer intended exactly. These most constrained lines have specified parametric relations to other line segments and those relations, if altered, will compromise the designer's intentions. Conversely, the lowest level of the hierarchy, which contains the least constrained line segments, only implies a specific connectivity between line segments, necessitating a vaster search. The parametric recognition of curved line shapes used a two-step approach that first performs shape matching with an equivalent straight-line shape then checks those transformations for matching with the actual curved lines. This approach has advantages characteristic polygons as well as emergent shapes.

Parametric shape grammar interpreter

A method of recognizing a first shape in a second shape. The method includes decomposing the first shape into at least one subshape belonging to one of a plurality of subshape groups, and searching the second shape for a parametric transformation of the subshape.

The present invention is directed to a method of, and apparatus for, solving a packing or layout problem by applying a pattern based search technique to an initial component configuration. By interactively working through a series of new or 'then current' configurations, an optimal component configuration is determined based on a predetermined criterion. According to one embodiment of the invention, at least one perturbation is introduced during the course of the search. The perturbation may be introduced by swapping component positions, increasing move size, or any other activity which breaks or interrupts the pattern of a normal pattern based search.

Method of optimizing component layout using a pattern based search

A method of optimizing a three-dimensional component layout in accordance with predetermined constraints is comprised of the steps of generating a plurality of models for each component. The plurality of models is arranged in a hierarchy from the model having the least resolution to the model having the most resolution. A starting layout is selected for the components. An iterative type of optimization routine is performed wherein each iteration is evaluated for satisfaction of the constraints using models for the components selected according to the level of resolution desired at that time in the iterative process. The optimization routine ends when a predetermined ending criterion is met.

Growing community of inventors

Pittsburgh, PA, United States of America

Jonathan Cagan