Embedded Gesture Detection Using Motion-Isolated Composite Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing gesture detection techniques face challenges in embedded systems due to the limitations of convolutional neural networks in handling series of time-related images and the resource-intensive nature of recurrent neural networks, which are not feasible in environments with limited computing resources, power supply, and connectivity.
Innovation Solution
A method involving an image sensor to capture a series of images, isolate motion using a processor, generate a composite image with a color gradient to indicate motion and time, and apply a convolutional neural network for gesture detection, reducing resource consumption and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If recurrent neural networks are used for gesture detection, then detection accuracy is improved, but resource consumption and system complexity increase significantly
Solution Approach 1:
The patent segments the gesture detection task into two distinct stages: (1) motion isolation stage that extracts motion information from video frames, and (2) gesture classification stage that uses a simplified neural network. This segmentation allows each stage to be optimized independently, reducing overall system complexity while maintaining accuracy.
Solution Approach 2:
The patent extracts only the essential motion information from video frames using motion isolation techniques, removing redundant visual data. This extraction process reduces the input data dimensionality significantly, allowing simpler neural networks to achieve the same detection accuracy without processing unnecessary information.
2Measurement precision
If recurrent neural networks are used for gesture detection, then detection accuracy is improved, but processor and power usage increase
Solution Approach 1:
The patent extracts only the essential motion information from video frames using motion isolation techniques, removing redundant visual data. This extraction process reduces the input data dimensionality significantly, allowing simpler neural networks to achieve the same detection accuracy without processing unnecessary information.
Solution Approach 2:
The patent replaces expensive, power-intensive recurrent neural networks with a more economical approach using motion isolation combined with simpler classification algorithms. This substitution maintains detection functionality while significantly reducing computational overhead and power consumption suitable for embedded systems.
3Device complexity
If conventional gesture detection techniques are used, then implementation is simpler, but resource consumption increases in embedded systems
Solution Approach 1:
The patent segments the gesture detection task into two distinct stages: (1) motion isolation stage that extracts motion information from video frames, and (2) gesture classification stage that uses a simplified neural network. This segmentation allows each stage to be optimized independently, reducing overall system complexity while maintaining accuracy.
Solution Approach 2:
The patent performs motion isolation as a preliminary action before gesture classification. By pre-processing the video data to extract only motion information, the system reduces the computational burden on subsequent processing stages, thereby lowering overall resource consumption while maintaining detection accuracy.
Data Source
AI summary
Disclosed herein are systems, devices, and processes for gesture detection. A method includes capturing a series of images. The method includes generating motion isolation information based on the series of images. The method includes generating a composite image based on the motion isolation information. The method includes determining a gesture based on the composite image. The processes described herein may include the use of convolutional neural networks on a series of time-related images to perform gesture detection on embedded systems or devices.