18100419. PERFORMING MULTIPLE SEGMENTATION TASKS simplified abstract (Adobe Inc.)

From WikiPatents
Jump to navigation Jump to search

PERFORMING MULTIPLE SEGMENTATION TASKS

Organization Name

Adobe Inc.

Inventor(s)

Jason Wen Yong Kuen of San Jose CA (US)

Zhe Lin of Clyde Hill WA (US)

Sukjun Hwang of Seoul (KR)

Jianming Zhang of Fremont CA (US)

Brian Lynn Price of Pleasant Grove UT (US)

PERFORMING MULTIPLE SEGMENTATION TASKS - A simplified explanation of the abstract

This abstract first appeared for US patent application 18100419 titled 'PERFORMING MULTIPLE SEGMENTATION TASKS

Simplified Explanation: The patent application describes a system that uses machine learning to segment objects in digital images for multiple tasks.

  • The computing device receives input data of a digital image.
  • Per-pixel embeddings are computed using a pixel decoder of a machine learning model.
  • Output embeddings are generated using a transformer decoder based on the per-pixel embeddings and input embeddings for two segmentation tasks.
  • The system outputs two digital images, each depicting the object segmented based on a different segmentation task.

Key Features and Innovation:

  • Utilizes machine learning models for image segmentation tasks.
  • Computes per-pixel embeddings for accurate segmentation.
  • Generates output embeddings for multiple segmentation tasks.
  • Outputs segmented digital images based on different segmentation tasks.

Potential Applications:

  • Image processing and editing software.
  • Medical imaging for identifying and analyzing specific areas.
  • Autonomous vehicles for object detection and segmentation.

Problems Solved:

  • Efficient and accurate segmentation of objects in digital images.
  • Handling multiple segmentation tasks simultaneously.
  • Enhancing the capabilities of image processing systems.

Benefits:

  • Improved accuracy in object segmentation.
  • Streamlined workflow for handling multiple segmentation tasks.
  • Enhanced performance of image analysis systems.

Commercial Applications:

  • "Machine Learning-Based System for Multi-Task Image Segmentation"
  • This technology can be applied in industries such as healthcare, autonomous vehicles, and digital imaging software.
  • Market implications include improved efficiency and accuracy in image segmentation tasks.

Questions about Multi-Task Image Segmentation: 1. How does the system handle different types of objects in the same image for segmentation tasks? 2. What are the potential limitations of using machine learning models for image segmentation tasks?

Frequently Updated Research: Ongoing research in the field of machine learning and computer vision may lead to advancements in multi-task image segmentation techniques. Stay updated on conferences and publications in these areas for the latest developments.


Original Abstract Submitted

In implementations of systems for performing multiple segmentation tasks, a computing device implements a segment system to receive input data describing a digital image depicting an object. The segment system computes per-pixel embeddings for the digital image using a pixel decoder of a machine learning model. Output embeddings are generated using a transformer decoder of the machine learning model based on the per-pixel embeddings for the digital image, input embeddings for a first segmentation task and input embeddings for a second segmentation task. The segment system outputs a first digital image and a second digital image. The first digital image depicts the object segmented based on the first segmentation task and the second digital image depicts the object segmented based on the second segmentation task.

Retrieved from "http://wikipatents.org/index.php?title=18100419._PERFORMING_MULTIPLE_SEGMENTATION_TASKS_simplified_abstract_(Adobe_Inc.)&oldid=206627"