Siamese Gesture Recognition for Low-Data Personalized Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current gesture recognition systems require a large amount of training data to accurately recognize personalized gestures, making it inconvenient for users to define new gestures.
Innovation Solution
A siamese neural network-based system where two neural network units share weights, calculate features from sensor signals, and adjust weights based on similarity analysis to reduce the need for extensive training data, allowing users to easily define personalized gestures with minimal training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a traditional neural network is used for gesture recognition, then the recognition accuracy can be ensured, but a large amount of training data is required which makes it inconvenient for users to define personalized gestures
Solution Approach 1:
The patent divides the training process into two distinct phases: pre-training phase using general gesture data to build foundational features, and fine-tuning phase using user-specific data to adapt to personalized gestures. This segmentation allows the system to achieve high recognition accuracy while requiring minimal user training data (only a few examples per gesture).
Solution Approach 2:
The system performs preliminary training on general gesture data before user-specific customization. The pre-trained neural network already possesses general gesture recognition capabilities, so when users want to define personalized gestures, they only need to provide a small number of training examples for quick adaptation, rather than training from scratch with large datasets.
2Reliability
If more training data is collected for personalized gesture definition, then the recognition accuracy improves, but the time and complexity of the training process increases
Solution Approach 1:
The training process is segmented into efficient phases: pre-training on general data (performed once) and quick fine-tuning on user data (performed per user). This segmentation enables the system to achieve high accuracy without requiring extensive training time from users, as the heavy lifting is done during pre-training.
Solution Approach 2:
The system changes training parameters dynamically: using large datasets with standard learning rates during pre-training, then switching to smaller datasets with adjusted learning rates during fine-tuning. This parameter adaptation allows rapid convergence with minimal user training data and time investment.
Data Source
AI summary
A gesture recognition system using siamese neural network executes a gesture recognition method. The gesture recognition method includes steps of: receiving a first training signal to calculate a first feature; receiving a second training signal to calculate a second feature; determining a distance between the first feature and the second feature in a feature space; adjusting the distance between the first feature and the second feature in feature space according to a predetermined parameter. Two neural networks are used to generate the first feature and the second feature, and determine the distance between the first feature and the second feature in the feature space for training the neural networks. Therefore, the gesture recognition system does not need a big amount of data to train one neural network for classifying a sensing signal. A user may easily define a new personalized gesture.