Salesforce, Inc.

Salesforce Com, Inc.

Mustbin, Inc.

Michael Reynolds Sollami

Jessica Lundin

Brian J Lonsdorf

David J Woodward

Owen Winne Schoppe

Alan Martin Ross

Sönke Rohde

Yang Zhang

Amir Hossein Raffiee

Preston C Shimer

Satyender Mahajan

Scott Prusinoski

Aashish Jain

Scott Bokma

Brian J Shin

Nathan A Mackey

Chetan Ramaiah

Xing Han

Xiangyu Peng

David James Woodward

Techniques are disclosed for automatically converting a layout image to a text-based representation. In the disclosed techniques, a server computer system receives a layout image that includes a plurality of portions representing a plurality of user interface (UI) elements included in a UI design. The server computer system transforms, via executed of a trained residual neural network (ResNet), the layout image to a text-based representation of the layout image that specifies coordinates of bounding regions of the plurality of UI elements included in the UI design, where the text-based representation is usable to generate program code executable to render the UI design. The disclosed techniques may advantageously automate one or more portions of a UI design process and, as a result save time and computing resources via the execution of an image to text-based conversion ResNet machine learning model.

Automatic image conversion

Systems, devices, and techniques are disclosed for visual text summary generation. An input text may be received. Keywords may be extracted from the input text. Representative keywords may be generated from the keywords. A graph representation of the representative keywords may be generated. Images associated with the representative keywords may be received. A visual-representation style may be selected based on the graph representation of the representative keywords. The images associated with the representative keywords may be arranged according to the selected visual-representation style and the graph representation of the representative keywords.

Visual text summary generation

Systems, devices, and techniques are disclosed for automatic product description generation. A first set of features including labels including words may be generated from an image using a first feature extraction model. A second set of features including labels including words may be generated from the image using a second feature extraction model. A text description of a product depicted in the image may be generated by inputting the image and metadata for the image to a description generating model. The text description may include words. Each of the words may be generated by assigning probabilities to candidate words, boosting the assigned probabilities of candidate words that are similar to words of labels of the first set of features or words of labels of the second set of features, and selecting one of the candidate words based on the assigned probabilities after the boosting as a word of the text description.

Automatic product description generation

A method receives an input file. The input file includes a plurality of entries where different entries are associated with a different characteristic. A selection of a guide for an entry is determined where a guide includes guidelines for formatting content in an entry in the plurality of entries. The method performs analyzing an image from the entry to generate image variations. Then, the method analyzes the image and the image variations to place text items from the entry in the image and the image variations where the text items in the image and the image variations are formatted based on the guidelines in the guide for the entry. A plurality of design variants is generated for the entry based on the image and the image variations that include text items.

Region-specific content creation

Systems and methods are provided for determining a first upper bound of a x-direction frame margin of a downsampled image and a second upper bound for a y-direction frame margin of the downsampled image. A total value for each row and column of a weighted grayscale image formed from the downsampled image may be determined. An outlier filtering may be applied to the determined total values for each row and column of the weighted grayscale image. A peak of pairwise ratios of consecutive values from left and right sides of the weighted grayscale image may be used to determine an x-margin, and a peak of pairwise ratios of consecutive values from top and bottom sides of the weighted grayscale image may be used to determine a y-margin. Content may be applied to the image based on the determined x-margin and y-margin values to form a composite image.

Systems and methods of determining margins of an image for content insertion to form a composite image

Techniques are disclosed for automatically generating new content using a trained 1-to-N generative adversarial network (GAN) model. In disclosed techniques, a computer system receives, from a computing device, a request for newly-generated content, where the request includes current content. The computer system automatically generates, using the trained 1-to-N GAN model, N different versions of new content, where a given version of new content is automatically generated based on the current content and one of N different style codes, where the value of N is at least two. After generating the N different versions of new content, the computer system transmits them to the computing device. The disclosed techniques may advantageously automate a content generation process, thereby saving time and computing resources via execution of the 1-to-N GAN machine learning model.

One-to-many automatic content generation

Disclosed herein are system, method and computer readable storage medium for detecting space suitable for overlaying media content onto an image. The system receives a candidate image which may be an image or a video frame. The candidate image is then input into a neural network. The neural network may output coordinates and one or more dimensions representing one or more bounding boxes for inserting media content into the candidate image. The one or more bounding boxes may be transmitted with a request for a media content item to be displayed in a bounding box. In response to the request the media content item may be received, and the candidate image and the media content item overlaid on top of the candidate image within the bounding box may be displayed.

Neural network based detection of image space suitable for overlaying media content

Systems and methods are provided for receiving an image that includes a clothed person, determining a pose of the person in the image, and segmenting the image into one or more first fashion items. One or more second fashion items may be determined using a similarity search that searches at least one storage device communicatively coupled to the server based on the one or more first fashion items. At least one outfit proposal may be generated based on the one or more second fashion items. Image re-stylization of corresponding portions of the image may be performed, including the clothed person to generate recommended outfit images based on the at least one outfit proposal. The generated outfit images may be transmitted for display.

Systems and methods of image-based neural network apparel recommendation

A method for data processing is described. The method includes receiving, from a client of a web service, a request for a web component included in a web page supported by the web service, where the web component is associated with a data structure that defines design characteristics of the web component. The method further includes determining configuration parameters that define stylistic characteristics of the web component based on data attributes of the web page and the data structure associated with the web component, where the configuration parameters are determined using a model trained on a corpus of web pages supported by the web service. The method further includes transmitting, to the client and in response to the request, an indication of the configuration parameters for the web component, where the indication is configured to display the web component according to the stylistic characteristics defined by the configuration parameters.

Techniques for automatically configuring web components

An online system generates a set of content item variations for a reference content item that include different styles of text for the content item. The different styles of text are generated by applying machine-learned style transfer models, for example, neural network based models to reference text of the reference content item. The text variations retain the textual content of the reference text but are synthesized with different styles. The online system can provide the content item variations to users on an online experimental platform to collect user interaction information that may indicate how users respond to different styles of text. The online system or the content providers can effectively target users with content items that include the style of text the users respond to based on the collected information.

Machine-learning based generation of text style variations for digital content items

Systems and methods are provided for receiving at least a first query color, and searching an electronic catalog including a plurality of product images for the first query color to determine a similarity measure between the first query color and a product image of a plurality of product images. The similarity measure may be determined by determining a Euclidean distance between values in a three-dimensional color space for the first query color and a target color of the product image, and determining the similarity measure between the query color and the product image by determining a sum of the similarity measures from all target colors on the product image, weighted by the coverage of each target color. The search results may be transmitted based on the searching of the electronic catalog including the plurality of product images for the first query color.

Systems and methods of multicolor search of images

Disclosed herein are system, method and computer readable storage medium for detecting space suitable for overlaying media content onto an image. The system receives an image which may be an image or a video frame. The image is processed using a number of image processing techniques in order to automatically propose spaces for inserting media content onto the image. The proposed spaces may then be further analyzed using a heuristics-based approach to select bounding boxes for inserting media content. Subsequently, one or more media content items may be selected for insertion onto the bounding boxes on the image. The system may then cause a display of the image with the selected media content item overlaid onto the image within the selected bounding boxes.

Heuristics-based detection of image space suitable for overlaying media content

A textual properties model is used to infer values for certain font properties of interest given certain text-related data, such as rendered text images. The model may be used for numerous purposes, such as aiding with document layout, identifying font families that are similar to a given font families, and generating new font families with specific desired properties. In some embodiments, the model is trained from a combination of synthetic data that is labeled with values for the font properties of interest, and partially-labeled data from existing 'real-world' documents.

Machine learning-based inference of granular font properties

A server of a generative adversarial network (GAN) for color selection; generates a training set of color palettes. A color palette generator of the server generates a first set of color palettes based on the training set of color palettes. The first set of color palettes may be compared with a reference set of color palettes to predict a curated set of color palettes. Colors from the curated set of color palettes may be removed that are within a predetermined distance from one another in a color space. The GAN may be validated by performing cluster analysis to determine outlier latent dimensions to be changed for the color selection by the GAN. Proposed color palettes may be generated based on the GAN to be displayed on a display device.

Systems and methods of generating color palettes with a generative adversarial network

Systems, device and techniques are disclosed for training data generation for visual search model training. A catalog including catalog entries which may include images of an item and data about the item may be received. Labels may be applied to the images of the items based on the data about the items. The images of the items may be sorted into clusters using cluster analysis on the labels. Each cluster may include labels as categories of the cluster. Additional images may be received based on searching for the categories. Generative adversarial network (GAN) training data sets may be generated from the images of the items, the additional images, and the categories. GANs may be trained with the GAN training data sets. The GANs may generate images including images of generated items, which may be replaced with images of items from the catalog entries to create feature model training images.

Growing community of inventors

Cambridge, MA, United States of America

Michael Reynolds Sollami