18053646. GENERATING IMAGE MATTES WITHOUT TRIMAP SEGMENETATIONS VIA A MULTI-BRANCH NEURAL NETWORK simplified abstract (ADOBE INC.)
Contents
- 1 GENERATING IMAGE MATTES WITHOUT TRIMAP SEGMENETATIONS VIA A MULTI-BRANCH NEURAL NETWORK
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 GENERATING IMAGE MATTES WITHOUT TRIMAP SEGMENETATIONS VIA A MULTI-BRANCH NEURAL NETWORK - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Potential Applications
- 1.6 Problems Solved
- 1.7 Benefits
- 1.8 Potential Commercial Applications
- 1.9 Possible Prior Art
- 1.10 Unanswered Questions
- 1.11 Original Abstract Submitted
GENERATING IMAGE MATTES WITHOUT TRIMAP SEGMENETATIONS VIA A MULTI-BRANCH NEURAL NETWORK
Organization Name
Inventor(s)
Zichuan Liu of San Jose CA (US)
Xin Lu of Mountain View CA (US)
GENERATING IMAGE MATTES WITHOUT TRIMAP SEGMENETATIONS VIA A MULTI-BRANCH NEURAL NETWORK - A simplified explanation of the abstract
This abstract first appeared for US patent application 18053646 titled 'GENERATING IMAGE MATTES WITHOUT TRIMAP SEGMENETATIONS VIA A MULTI-BRANCH NEURAL NETWORK
Simplified Explanation
The disclosed patent application describes a system for generating image mattes for detected objects in digital images without trimap segmentation using a multi-branch neural network approach.
- The system utilizes a generative neural network with multiple branches to extract a coarse semantic mask, a detail mask, and then fuse them to generate an image matte.
- A refinement neural network is also used to further enhance the image matte generated by the generative neural network.
Potential Applications
This technology can be applied in various fields such as image editing, computer vision, and augmented reality for precise object segmentation in digital images.
Problems Solved
This technology solves the problem of generating accurate image mattes for detected objects without the need for manual trimap segmentation, saving time and effort in image editing tasks.
Benefits
The benefits of this technology include improved efficiency in object segmentation, enhanced image editing capabilities, and the ability to generate high-quality image mattes for various applications.
Potential Commercial Applications
Potential commercial applications of this technology include image editing software, photography tools, and computer vision systems for automated object segmentation in images.
Possible Prior Art
One possible prior art for this technology could be traditional image segmentation techniques that require manual input or trimap segmentation for object extraction in digital images.
Unanswered Questions
How does this technology compare to existing image segmentation methods in terms of accuracy and efficiency?
The article does not provide a direct comparison with existing image segmentation methods to evaluate the accuracy and efficiency of this technology.
What are the limitations of the multi-branch neural network approach in generating image mattes for complex objects or scenes?
The article does not address the potential limitations of the multi-branch neural network approach when dealing with complex objects or scenes in digital images.
Original Abstract Submitted
Methods, systems, and non-transitory computer readable storage media are disclosed for generating image mattes for detected objects in digital images without trimap segmentation via a multi-branch neural network. The disclosed system utilizes a first neural network branch of a generative neural network to extract a coarse semantic mask from a digital image. The disclosed system utilizes a second neural network branch of the generative neural network to extract a detail mask based on the coarse semantic mask. Additionally, the disclosed system utilizes a third neural network branch of the generative neural network to fuse the coarse semantic mask and the detail mask to generate an image matte. In one or more embodiments, the disclosed system also utilizes a refinement neural network to generate a final image matte by refining selected portions of the image matte generated by the generative neural network.
