Description:The present invention relates to an Artificial Intelligence (AI) powered smart eyeglass that uses advanced deep learning algorithms to extract real-world visual information and convert it into a knowledge graph. This knowledge graph will be used for many further applications, such as hidden relationship identification and other memory remembrance applications. Smart eyeglasses combine computer vision and deep-learning algorithms to provide fast and accurate results. , Claims:An AI-powered smart eyeglass generates scene graph knowledge for clinical surveillance comprising:
- A plurality of cameras collects the visual data and audio data of the real time scene in OR.
- The collected image and audio data will be stored and pre-processed using the preprocessing modules.
- The feature extraction modules extract the required features by using the pre-processed image and audio.
- Based on the image features the object detection model will predict the labels for the objects.
- By using the object labels and audio features the relationship model will predict the relationship between the objects.
- In the graph generator module, based on the objects and their respective relationships the graph will generated.
- The generated graph will be updated with the existing graph in the graph bank by using evo graph model.
- Based on the generated graph will be analyzed used for further analysis and to-do action with it. These results will be displayed or announced by the inbuilt speaker and as well as the user can send an action for the robots.
The system of claim 1, wherein the image includes a camera and audio device that captures the image of the food dish.
The system of claim 1, wherein the pre-processor modules cleans, filters, and transforms the image and audio with the suitable requirement for the feature extraction model.
The system of claim 1, wherein the feature extraction models it detect the relevant and required features from the processed image and audio.
The system of claim 1, wherein the object detection algorithm predicts the object labels of the scene using depth estimation networks.
The system of claim 1, wherein the relationship model predicts the relationship based on the object labels and as well as the audio features.
The system of claim 1, wherein the graph generation module generates the graph based on the objects and relationships.
The system of claim 1, wherein the knowledge graph generator updates the graph with the existing knowledge graph with the help of graph learning algorithms.
The system of claim 2, further comprises the use language model for getting the results from the knowledge graph with respect to the input queries.
A computer program product for implementing an AI-powered smart eyeglass generates scene graph knowledge for clinical surveillance system computer program product comprised:
a. Computer code for collecting data of the real time scene using a camera device.
b. Computer code for storing and processing the data using a pre-processor module.
c. Computer code for feature extraction of the pre-processed image and audio using a deep learning algorithm.
d. Computer code for object classification and relationship prediction by using deep learning algorithms.
e. Computer code for graph generation deep learning algorithms.
f. Computer Code for graph storage, update by deep learning algorithm and retrieval of the generated graph in the graph database.
g. Computer Code for projecting and assigning the query results in the monitor, speaker, or Todo robot action scripts.