Footprint system capable of obtaining a human body living state
A technology of living status and footprint, applied in the field of footprint system, can solve problems such as single function, lack of living status, monitoring, etc.
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0096] This embodiment provides a footprint system capable of obtaining human living conditions, including:
[0097] 1. Mental state assessment module; input of the assessment module: real-time footprint image sequence with time tags (for the same person, it is impossible to have a short-term multi-sensor image collection, so it is considered that the image source of the same person sequence is unified ), including two types of graphs with stress response and non-stress response, evaluating the objective information of the object (height, age, weight, gender entered by the user himself).
[0098] a) Objective information calibration based on CNN network:
[0099] i. Preprocessing: the integration of sequence diagrams, transforming real-time footprints into complete footprints;
[0100] ii. Use the pre-trained model to determine height, age, and weight (gender is not required). When the difference between the determined age and the input age exceeds 1 year, the input informati...
Embodiment 2
[0186] This embodiment provides a pre-trained deep learning age network model B, specifically:
[0187] S1: Acquire barefoot or socks footprint image data, and preprocess the image data;
[0188] S2: Create a barefoot image dataset;
[0189] S3: Data training and feature extraction, the improved AlexNet network is used here, the specific structure is as follows:
[0190] (1) Initial network:
[0191] Network composition: 4 layers of convolutional layers, 2 layers of pooling layers, and 2 layers of fully connected layers;
[0192] Internet connection:
[0193] conv1+pooling1+relu→conv2+pooling2+relu→conv3+relu→conv4+relu→fc5→fc6
[0194] Among them, conv represents the convolutional layer, pooling represents the pooling layer, fc represents the fully connected layer, and relu represents the activation function;
[0195] (2) Adjust the network through training and verification results: After using the initial network for a complete training, test with the verification data,...
Embodiment 3
[0199] This embodiment provides a pre-trained deep learning weight network model C, specifically:
[0200] S1: Acquire barefoot or socks footprint image data, and preprocess the image data;
[0201] S2: Create a barefoot image dataset;
[0202] 1) Divide the preprocessed barefoot image dataset into two parts:
[0203] (1) Training set: used for the training process of deep learning, each barefoot footprint data sample has subordinate weight information, and this weight information is the label of the barefoot or socks footprint;
[0204] (2) Verification set: used to verify the quality of deep learning results. Each barefoot or sock footprint data sample has subordinate weight information, but the validation set does not participate in training, it is only used to measure the accuracy of weight determination
[0205] 2) Among them, the data requirements of each part:
[0206] (1) The data dimension of the verification set shall not be higher than that of the training set d...
PUM
Abstract
Description
Claims
Application Information
- R&D Engineer
- R&D Manager
- IP Professional
- Industry Leading Data Capabilities
- Powerful AI technology
- Patent DNA Extraction
Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.
© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap