Robust training of large-scale object detectors with a noisy dataset

US11126890B2 · kind B2 · utility

Assignee

• Adobe Inc. · US

Inventors

• Zhe Lin · Fremont, US
• Mingyang Ling · San Francisco, US
• Jianming Zhang · Campbell, US
• Jason Wen Yong Kuen · Singapore, SG
• Federico Perazzi · San Francisco, US
• Brett Butterfield · San Francisco, US
• Baldo Faieta · San Francisco, US

Key dates

• Filing date: Apr 18, 2019
• Grant date: Sep 21, 2021
• Expiry date: Dec 30, 2039

Classification

• Technology area (CPC G): Physics
• CPC primary: G06V10/82
• WIPO field: Computer technology
• WIPO sector: Electrical engineering

Abstract

Systems and methods are described for object detection within a digital image using a hierarchical softmax function. The method may include applying a first softmax function of a softmax hierarchy on a digital image based on a first set of object classes that are children of a root node of a class hierarchy, then apply a second (and subsequent) softmax functions to the digital image based on a second (and subsequent) set of object classes, where the second (and subsequent) object classes are children nodes of an object class from the first (or parent) object classes. The methods may then include generating an object recognition output using a convolutional neural network (CNN) based at least in part on applying the first and second (and subsequent) softmax functions. In some cases, the hierarchical softmax function is the loss function for the CNN.