Adobe, Inc.

Ebay, Incorporated

Daniel Miranda

Zhe Lin

David Goldberg

Midhun Harikumar

Akhilesh Kumar

Kshitiz Garg

Prathyusha Senthil Kumar

Xiaozhen Xue

Jianming Zhang

Baldo Antonio Faieta

Shabnam Ghadar

Jayant Kumar

Saeid Motiian

Nicolas Huynh Thien

Pramod Srinivasan

Kannan Iyer

Rohith Mohan Dodle

Gaurav Kukal

Vipul Dalal

Vera Lychagina

Mina Doroudi

Archit Kalra

Anthony Robert Bell

Amit Jaiswal

Prethebha Chandrasegaran

Charles R McKinney

Ishita Kamal Khan

This disclosure describes one or more implementations of a video inference system that utilizes machine-learning models to efficiently and flexibly process digital videos utilizing various improved video inference architectures. For example, the video inference system provides a framework for improving digital video processing by increasing the efficiency of both central processing units (CPUs) and graphics processing units (GPUs). In one example, the video inference system utilizes a first video inference architecture to reduce the number of computing resources needed to inference digital videos by analyzing multiple digital videos utilizing sets of CPU/GPU containers along with parallel pipeline processing. In a further example, the video inference system utilizes a second video inference architecture that facilitates multiple CPUs to preprocess multiple digital videos in parallel as well as a GPU to continuously, sequentially, and efficiently inference each of the digital videos.

Efficiently inferencing digital videos utilizing machine-learning models

Increasing efficiency of inferencing digital videos utilizing machine-learning models

The present disclosure relates to an image merging system that automatically and seamlessly detects and merges missing people for a set of digital images into a composite group photo. For instance, the image merging system utilizes a number of models and operations to automatically analyze multiple digital images to identify a missing person from a base image, segment the missing person from the second image, and generate a composite group photo by merging the segmented image of the missing person into the base image. In this manner, the image merging system automatically creates merged group photos that appear natural and realistic.

Automatically merging people and objects from multiple digital images to generate a composite digital image

Various methods and systems for providing query result items using an item title demand model are provided. A query is received at a search engine. Based on receiving the query, an item title demand engine is accessed. The item title demand engine operates based on an item title demand model which uses token weights, representing skip probabilities of tokens in item titles, to determine title scores for result item titles for corresponding queries. Based on accessing the item title demand engine, one or more result item titles for the query are identified from items in an item database. An identified result item title is identified based on a title score determined using the item title demand model and a highest skip probability of a token in the result item title. The one or more result item titles are communicated to cause display of the one or more result item titles.

Search system for providing query result items based on highest skip probabilities of item title tokens

A query for one or more resources is received. One or more tokens associated with the query is identified based on running the query through a learning model. The one or more tokens correspond to one or more terms that the query shares context similarity to based on a history of user selections. One or more search result candidates are scored based at least on the context similarity between the one or more tokens and the query.

Comprehensive search engine scoring and modeling of user relevance

Object search techniques for digital images are described. In the techniques described herein, semantic features are extracted on a per-object basis form a digital image. This supports location of objects within digital images and is not limited to semantic features of an entirety of the digital image as involved in conventional image similarity search techniques. This may be combined with indications a location of the object globally with respect to the digital image through use of a global segmentation mask, use of a local segmentation mask to capture post and characteristics of the object itself, and so on.

Object search in digital images

A search engine is configured to return increased diversity results based on past user interactions with search results. For a given query, historical data is analyzed to generate an item score describing a past quantity of users that navigated to a given page of an item. The historical data can further be used to generate a category score describing a past quantity of users that navigated to a given category of items. The category of items can be analyzed to generate a diversity score describing their diversity with respect to each other. Results for the given query can be arranged using items scores, category scores, and diversity scores.

Dissimilar but relevant search engine results

Growing community of inventors

San Jose, CA, United States of America

Daniel Miranda