Data Capturing Device Using Server-Deployed Machine Learning Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional smart cameras require complex hardware designs and high costs due to their need for independent computing and fixed firmware, limiting flexibility in function and cost-effectiveness.
Innovation Solution
A data capturing device and system that utilizes a sensor, first and second processing circuits, and communication chips to run machine learning models and libraries, with a computing device deploying these components to the data capturing device, allowing for flexible and cost-effective operation using simple hardware.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional smart cameras perform computing independently with burned-in firmware, then they can provide fixed functions, but the hardware design becomes complex and costs increase
Solution Approach 1:
The patent extracts the computing functions and firmware from the camera hardware, moving them to an external server. The camera retains only basic data capture capabilities, while complex processing is performed externally, thereby reducing hardware complexity while maintaining functionality
Solution Approach 2:
The server provides a universal platform that can serve multiple cameras with different functions through software deployment. A single camera hardware design can be configured for various purposes (surveillance, analysis, control) by deploying different machine learning models and services on the server, achieving multi-functionality without hardware changes
2Adaptability or versatility
If conventional smart cameras use burned-in firmware, then functions are fixed, but flexibility to change functions is lost
Solution Approach 1:
The system transitions from static burned-in firmware to dynamic software that can be deployed, updated, and modified on the server. Machine learning models and service functions can be changed remotely without hardware modifications, enabling the system to adapt to new requirements while simplifying the manufacturing process
Solution Approach 2:
The server acts as an intermediary between the camera hardware and the functional software. It manages the deployment, updates, and execution of machine learning models and services, providing a flexible interface that allows function changes without complicating the camera manufacturing process
3Productivity
If conventional smart cameras perform large amounts of calculations independently, then they can process data locally, but costs increase due to complex hardware requirements
Solution Approach 1:
The system segments the data processing function into two parts: data capture by the camera and data processing by the server. This division allows the camera to use simple hardware while the server handles complex calculations, achieving high productivity without increasing camera hardware complexity
Solution Approach 2:
The patent uses machine learning models that can be copied and deployed to the server, enabling complex data processing capabilities without requiring complex hardware in the camera. The same processing logic can be replicated across multiple servers or instances, maintaining productivity while keeping individual device complexity low
Data Source
AI summary
A data capturing device and a data calculation system and method are provided. The data capturing device transmits sensing data to a computing device, and receives a machine learning model and a library corresponding to a current scene from the computing device. The data capturing device runs the machine learning model to capture feature data from the sensing data, runs the library to convert a requirement into a service task, and then transmits the feature data and the service task.