18170336. TRANSFORMER-BASED IMAGE SEGMENTATION ON MOBILE DEVICES simplified abstract (ADOBE INC.)
Contents
TRANSFORMER-BASED IMAGE SEGMENTATION ON MOBILE DEVICES
Organization Name
Inventor(s)
Jingyuan Liu of Santa Clara CA (US)
Qing Liu of Santa Clara CA (US)
Yuhong Wu of Sammamish WA (US)
TRANSFORMER-BASED IMAGE SEGMENTATION ON MOBILE DEVICES - A simplified explanation of the abstract
This abstract first appeared for US patent application 18170336 titled 'TRANSFORMER-BASED IMAGE SEGMENTATION ON MOBILE DEVICES
The present disclosure pertains to systems, computer-readable media, and methods for creating segmentation masks for digital visual media items using neural networks.
- Utilizes a neural network encoder to generate high-level features of the digital visual media item.
- Generates low-level features of the digital visual media item using the neural network encoder.
- Creates an initial segmentation mask of the digital visual media item from the low-level features with a neural network decoder.
- Refines the segmentation mask of the digital visual media item using the neural network decoder and the high-level features.
Potential Applications: - Image and video editing software - Object recognition and tracking systems - Augmented reality applications
Problems Solved: - Efficient and accurate segmentation of digital visual media items - Automation of segmentation mask generation process
Benefits: - Improved editing capabilities for digital visual media - Enhanced object recognition accuracy - Streamlined workflow for content creators
Commercial Applications: Title: Advanced Segmentation Mask Generation Technology for Digital Media Editing This technology can be utilized in various industries such as graphic design, video production, and computer vision for tasks like image segmentation, object removal, and background replacement.
Prior Art: Researchers can explore existing patents related to neural network-based image segmentation and object recognition techniques to understand the prior art in this field.
Frequently Updated Research: Stay updated on advancements in neural network algorithms for image processing and segmentation to enhance the efficiency and accuracy of segmentation mask generation.
Questions about Advanced Segmentation Mask Generation Technology: 1. How does this technology improve the efficiency of digital media editing processes? 2. What are the key differences between high-level and low-level features in the context of segmentation mask generation?
Original Abstract Submitted
The present disclosure relates to systems, non-transitory computer-readable media, and methods for generating segmentation masks for a digital visual media item. In particular, in one or more embodiments, the disclosed systems generate, utilizing a neural network encoder, high-level features of a digital visual media item. Further, the disclosed systems generate, utilizing the neural network encoder, low-level features of the digital visual media item. In some implementations, the disclosed systems generate, utilizing a neural network decoder, an initial segmentation mask of the digital visual media item from the low-level features. Moreover, the disclosed systems generate, utilizing the neural network decoder, a refined segmentation mask of the digital visual media item from the initial segmentation mask and the high-level features.
