Amazon Technologies, Inc.

A9.com, Inc.

Paulo Mendonca

Yifan Xing

Kunwar Yashraj Singh

Zhongwei Cheng

Rahul Tewari

Techniques for machine learning-based image super-resolution are described. A Dual Path Deep Back Projection Network can be used to enhance an input image. For example, the model may be trained to perform image super-resolution, remove artifacts, provide filtering or low light enhancement, etc. Classification may be performed on the resulting enhanced images to identify objects represented in the images. The model may be trained using a dataset that includes groups of images: an original image and an enhanced image. The model may use both residual and dense connectivity patterns between each successive back projection blocks to improve construction of a high-resolution output image from a low resolution input image. The enhanced images increase classification accuracy for input images having low image resolution.

DualPath Deep BackProjection Network for super-resolution

Techniques for improving a 2D to 3D image reconstruction network machine learning model are described. In some instances, this includes performing at least two transformations of a 3D model to generate at least two rotated 3D models, the at least two transformations to rotate the 3D model about an axis away from a viewing direction of the single 2D image; rendering the at least two rotated 3D models as rendered 2D images; and retraining a 2D to 3D image reconstruction network machine learning model using corresponding pairs of rotated 3D models and rendered 2D images.

Self-supervised bootstrap for single image 3-D reconstruction

Image data may be downloaded or otherwise obtained. An application executing on the computing device (or at least in communication with the computing device) can analyze the image data to recognize objects represented in the image data as well as determine locations or regions of the image that include the representations. Candidate regions of interest (e.g., bounding boxes) that include at least one detected object can be generated. A single region&#x2014;a representative region&#x2014;can be determined from the candidate regions based on user and/or application-defined criterion. The criteria may help ensure that the representative region includes statistical properties that at least satisfy threshold property levels, such as being a consistent estimator of the candidate regions. Notions of 'average' or 'median' regions can be applied to generate the representative region. The representative region can be associated with a confidence score, which can reflect inconsistency in the candidate regions, and therefore can be associated with a property of the candidate regions and not of the representative region alone.

Growing community of inventors

Seattle, WA, United States of America

Paulo Mendonca