Cross-View Gait Recognition via Deep Learning Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Prior gait recognition methods face significant challenges in achieving practical precision when dealing with severe cross-view problems, as they struggle to accurately match gait energy images from different angles.
Innovation Solution
A deep learning-based gait recognition method using a convolutional neural network to train a matching model by extracting gait energy images from multiple views, selecting positive and negative samples, and optimizing the model through error back propagation until convergence, enabling accurate cross-view gait recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional gait recognition methods are used to process cross-view gait recognition, then the system complexity remains low, but the recognition precision deteriorates significantly
Solution Approach 1:
The patent introduces Gait Energy Images (GEI) as an intermediary representation to bridge the gap between different view angles. GEI transforms temporal gait sequences into energy distribution maps that capture motion patterns while being more robust to view changes, serving as a mediator that enables cross-view matching without requiring complex view transformation models
Solution Approach 2:
The patent transforms the gait recognition problem from direct video sequence comparison to GEI-based energy distribution comparison. By changing the parameter space from temporal-spatial video frames to energy magnitude distributions, the system achieves better cross-view generalization while using a relatively simple nearest-neighbor classifier
2Measurement precision
If deep convolutional neural network is used to train matching model, then the gait recognition precision is improved, but the training time and computational resources increase
Solution Approach 1:
The patent performs preliminary action by pre-training the deep convolutional neural network on large-scale datasets to learn generalizable gait features. The pre-trained model serves as a foundation that can be fine-tuned with smaller datasets, reducing the training time required for specific applications while maintaining high recognition precision
Solution Approach 2:
The patent uses transfer learning by copying knowledge from pre-trained deep neural networks to the gait recognition task. Instead of training from scratch, the system adapts existing deep learning models to gait data, significantly reducing training time while preserving the high precision capabilities of deep networks
Data Source
AI summary
The present disclosure relates to a gait recognition method based on deep learning, which comprises recognizing an identity of a person in a video according to the gait thereof through dual-channel convolutional neural networks sharing weights by means of the strong learning capability of the deep learning convolutional neural network. Said method is quite robust to gait changes across a large view, which can effectively solve the problem of low precision in cross-view gait recognition existing with the prior art gait recognition technology. Said method can be widely used in scenarios having video monitors, such as security monitoring in airports and supermarkets, person recognition, criminal detection, etc.