Target Re-Identification Network With Multi-Branch Local Feature Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing target re-identification networks have low accuracy in target identification due to inadequate feature extraction and representation.
Innovation Solution
A target re-identification network is trained with multiple branches that extract n local feature information from different regions of an image, followed by parameter adjustment and fusion of feature information to enhance representation and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing training methods are used for target re-identification network, then the network can be trained and deployed, but the target identification accuracy is low
Solution Approach 1:
The target re-identification network is divided into multiple branches, where each branch extracts features from different regions of the input image. This segmentation allows the network to capture local features from various parts of the target simultaneously, improving identification accuracy by focusing on region-specific characteristics rather than processing the entire image as a single unit.
Solution Approach 2:
The network extracts n local feature information (where n > 3) from different regions, adding a spatial dimension to feature extraction. Instead of using a single global feature representation, the patent introduces multiple local feature vectors that capture information from different spatial locations, thereby enriching the feature space and improving discrimination capability.
2Loss of information
If multiple branches with n local feature information are used, then feature extraction granularity is enriched, but network complexity increases
Solution Approach 1:
The network is segmented into multiple independent branches, each responsible for extracting features from a specific region. This segmentation ensures that local features from different regions are captured without interference, reducing information loss while maintaining manageable complexity through modular architecture.
Solution Approach 2:
Multiple branches perform the same feature extraction function but on different regions of the input image. This multi-functionality allows the network to gather comprehensive feature information from various locations simultaneously, improving feature representation completeness without requiring fundamentally different processing mechanisms for each region.
Data Source
AI summary
A device for transferring micro-components and a method of manufacturing the device are provided. The device includes a substrate, a metal wire, and a plurality of silicon electrodes. The metal wire is formed on a flat surface of the substrate and includes a plurality of electrode driving units. The silicon electrodes are formed on a side of the metal wire opposing to the substrate. Each silicon electrode corresponds to each electrode driving unit and is driven by the electrode driving unit to pick or release each micro-component. According to the present application, the device may electrostatically adsorb a massive amount of micro-components to achieve transferring the massive amount of the micro-components, dramatically improving a transfer efficiency.