Behavior cross-domain identification model establishment and identification method and system based on CSI data
A technology for identifying models and establishing methods, applied in the field of behavior recognition, can solve problems such as behavior recognition in difficult target domains, and achieve the effect of strengthening the learning process, fast and high-precision model adaptation, and increasing available information
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0058] This embodiment discloses a method for establishing a behavioral cross-domain recognition model based on CSI data. The method is used to build a cross-domain recognition model, which mainly includes the following steps:
[0059] Step 1. Establish source domain behavior datasets and target domain behavior datasets, each dataset contains N types of behaviors, N≥1;
[0060] In this embodiment, one tester is set to perform 10 types of behavior actions in a medium-sized conference room, a small conference room, a small laboratory or an office, that is, N=10, and 20 types of behaviors of each type are collected in each scene. sample data, and the data of each category in each scene is recorded in the same long sequence signal data; two of the scenes are selected as the source domain and the target domain, and all the data in each domain are recorded as a data set, respectively. Be the source domain behavior dataset and the target domain behavior dataset. And the behavioral d...
Embodiment 2
[0082] This embodiment discloses a behavior cross-domain recognition model establishment system based on CSI data, the system includes a sample data construction module, a sample data segmentation module, a sample data re-representation module, a data pairing module, a twinning module, and a spatio-temporal feature distance calculation module and the model training module, where,
[0083] The sample data building block is used to collect source domain behavior data and target domain behavior data. Each dataset contains N types of behaviors, all source domain behavior data is saved to the source domain dataset, and all target domain behavior data is saved to the target domain dataset. In this embodiment, TP-LINK routers and laptops are used to collect sample data.
[0084]The sample data segmentation module is used to segment the data in the source domain dataset and the target domain dataset respectively to obtain all independent behavioral data fragments in the source domain...
Embodiment 3
[0095] This embodiment discloses a behavior cross-domain identification method based on CSI data, such as figure 1 As shown, the method specifically includes the following steps:
[0096] Step 1, collecting the behavior data of the test object, where the scene where the test object is located is the same scene as the target domain during the recognition model training.
[0097] Step 2: Use the method described in Step 2 in Embodiment 1 to process the behavior data of the test object to obtain re-represented behavior data fragments. In this embodiment, the remaining behavior data segments in the target domain data set in Embodiment 1 except the training samples are selected as the behavior data segments to be identified.
[0098] Step 3: pair the re-represented behavior data fragments with the re-represented behavior data fragments of each category in the source domain data set in Example 1 to obtain a data pair. In this embodiment, preferably, the re-represented behavior dat...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More