Panasonic intellectual property management co., ltd. (20240119717). PROCESSING METHOD AND PROCESSING DEVICE USING SAME simplified abstract
Contents
- 1 PROCESSING METHOD AND PROCESSING DEVICE USING SAME
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 PROCESSING METHOD AND PROCESSING DEVICE USING SAME - 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
PROCESSING METHOD AND PROCESSING DEVICE USING SAME
Organization Name
panasonic intellectual property management co., ltd.
Inventor(s)
PROCESSING METHOD AND PROCESSING DEVICE USING SAME - A simplified explanation of the abstract
This abstract first appeared for US patent application 20240119717 titled 'PROCESSING METHOD AND PROCESSING DEVICE USING SAME
Simplified Explanation
The patent application describes a system that combines the outputs of two neural networks to generate a final feature map that is larger than the original target image.
- The first processor executes processing of a first neural network on the target image and generates a first feature map that is 1/m x 1/n times as large as the target image.
- An enlarger then enlarges the first feature map by n times.
- The second processor executes processing of a second neural network on the target image and generates a second feature map that is 1/m times as large as the target image.
- The combiner combines the enlarged first feature map and the second feature map to produce the final output.
Potential Applications
This technology could be applied in image processing, object recognition, and computer vision applications.
Problems Solved
This technology solves the problem of generating high-resolution feature maps from neural networks that operate on smaller input images.
Benefits
The system allows for the generation of detailed and comprehensive feature maps by combining the outputs of multiple neural networks.
Potential Commercial Applications
One potential commercial application of this technology could be in the development of advanced image recognition systems for security, surveillance, and autonomous vehicles.
Possible Prior Art
Prior art may include similar systems that combine the outputs of multiple neural networks for image processing tasks.
Unanswered Questions
How does the system handle potential conflicts or discrepancies between the outputs of the two neural networks?
The system may include mechanisms to resolve conflicts or discrepancies between the outputs of the two neural networks, such as weighted averaging or thresholding.
What is the computational overhead of combining the outputs of multiple neural networks in real-time applications?
The computational overhead of combining the outputs of multiple neural networks in real-time applications may vary depending on the complexity of the networks and the size of the input images.
Original Abstract Submitted
a first processor executes processing of a first neural network on a target image to be processed and generates a first feature map having a size (1/m)�(1/n) times as large as the target image . an enlarger enlarges the first feature map generated in the first processor by n times. a second processor executes processing of a second neural network on the target image and generates a second feature map having a size (1/m) times as large as the target image . the combiner combines the first feature map enlarged by n times in the enlarger and the second feature map generated in the second processor
