Paypal, Inc.

Xiaodong Yu

Quan Jin Ferdinand Tang

Jiyi Zhang

Yuzhen Zhuo

Hewen Wang

Jiazheng Zhang

Shanshan Peng

Runmin Wen

Hong Qian

Zhe Chen

Kai Xie

Yanfei Dong

Yuan Deng

Shek Hei Wong

Quan Anh Nguyen

Hans Tananda

Long Phi Huynh

Methods and systems are presented for extracting categorizable information from an image using a graph that models data within the image. Upon receiving an image, a data extraction system identifies characters in the image. The data extraction system then generates bounding boxes that enclose adjacent characters that are related to each other in the image. The data extraction system also creates connections between the bounding boxes based on locations of the bounding boxes. A graph is generated based on the bounding boxes and the connections such that the graph can accurately represent the data in the image. The graph is provided to a graph neural network that is configured to analyze the graph and produce an output. The data extraction system may categorize the data in the image based on the output.

Content extraction based on graph modeling

Systems and/or techniques for facilitating image forgery detection via pixel-metadata consistency analysis are provided. In various embodiments, a system can receive an electronic image from a client device. In various cases, the system can obtain a pixel vector and/or an image metadata vector that correspond to the electronic image. In various aspects, the system can determine whether the electronic image is authentic or forged, based on analyzing the pixel vector and the image metadata vector via at least one machine learning model.

Image forgery detection via pixel-metadata consistency analysis

A system can comprise a processor that can facilitate performance of operations, comprising accessing a document comprising a plurality of text bounding boxes, wherein each respective text bounding box of the plurality of text bounding boxes comprises respective text, for each respective text bounding box, determining respective text bounding box coordinates and respective text bounding box input embeddings, based on the respective text bounding box coordinates, determining respective text bounding box positional encodings for each respective text bounding box, based on a transformer-based deep learning model applied to the respective text bounding box input embeddings, respective text bounding box coordinates, respective text bounding box positional encodings, and bias information representative of a modification to an attention weight of the transformer-based deep learning model, determining respective output embeddings for each respective text bounding box, and based on the respective output embeddings, generating respective bounding box labels for each respective bounding box.

Polar relative distance transformer

Systems and/or techniques for facilitating image forgery detection via headpose estimation may include a system that can receive a document from a client device. The system can identify, by executing a first trained machine learning model, an object that is depicted in the document. The system can determine, by executing a second trained machine learning model, a pose of the object. The system can determine, by executing a third trained machine learning model, whether the document is authentic or forged based on the pose of the object. The system can, in response to determining that the document is forged, transmit an unsuccessful validation message to the client device.

Image forgery detection via headpose estimation

Aspects of the present disclosure involve systems, methods, devices, and the like for generating an adversarially resistant model. In one embodiment, a novel architecture is presented that enables the identification of an image that has been adversarially attacked. The system and method used in the identification introduce the use of a denoising module used to reconstruct the original image from the modified image received. Then, further to the reconstruction, an adversarially trained model is used to make a prediction using at least a determination of a loss that may exist between the original image and the denoised image.

System and method for counteracting adversarial attacks

Aspects of the present disclosure involve systems, methods, devices, and the like for user identification using Artificial Intelligence, Machine Learning, and data analytics. In one embodiment, a verification system and method is introduced that can provide user authentication using parallel modeling for face identification. The verification system used includes a face identification module for use in the identification and verification using parallel processing of a received image with a claimed identity. The parallel processing includes an ensemble of machine learning models processed in parallel for optimal performance.

Ensemble method for face recognition deep learning models

A system can merge text bounding boxes such as Optical Character Recognition (OCR) bounding boxes. A document can comprise a plurality of the text bounding boxes. Distance thresholds between text bounding boxes can be utilized for comparison against a distance threshold. Distance thresholds can vary depending on context information associated with the document. In response to a determination that text bounding boxes satisfy the distance threshold, the text bounding boxes can be assigned to a bounding box group.

Efficient bounding box merging

Techniques are disclosed relating to determining whether document objects included in an image correspond to known document types. In some embodiments, a computing system maintains information specifying a set of known document types. In some embodiments, the computing system receives an image that includes objects. In some embodiments, the computing system analyzes, using a first neural network, the image to identify a document object and location information specifying a location of the document object within the image. In some embodiments, the computing system determines, using a second neural network, whether the document object within the image corresponds to a document type specified in the set of known document types, where the determining is performed based on the location information of the document object. In some embodiments, disclosed techniques may assist in automatically extracting information from documents, which in turn may advantageously decrease processing time for onboarding new customers.

Growing community of inventors

Singapore, Singapore

Xiaodong Yu