An intelligent evaluation system based on an internet of things terminal and a cloud content analysis engine and a dual-track service method thereof

Abstract

The application discloses an intelligent evaluation system based on a special content collection terminal and a cloud AI analysis engine and a double-track service method thereof. The system comprises a content collection terminal, a cloud service platform and a user terminal. The core innovation lies in that the content collection terminal is an independent hardware device separated from the user terminal, supports WiFi and 4G / 5G cellular dual-mode communication, completes content collection and cloud uploading through one-key operation, automatically switches to a cellular network when WiFi is unavailable, and the collection terminal is physically separated from the viewing terminal; a device code management module adopts an anti-obfuscation character exclusion rule to generate device codes in batches and supports batch tracking management; a subscription and permission management module implements a double-track service strategy - device-activated users automatically obtain a preset validity period, and non-device-registered users obtain a trial period and pay by period; a cloud analysis engine adopts a multi-stage pipeline architecture, comprises four stages of content preprocessing, content recognition, category determination and item-by-item evaluation, and is specifically implemented as an AI correction engine when applied to a homework correction scene; an Internet of Things traffic management module realizes monitoring, quota management, operator docking and revenue sharing of cellular traffic, and integrates traffic cost into the subscription system. The application further comprises a compatible processing mechanism for historical lifelong users. The system solves the problems of the prior art that the collection terminal and the entertainment terminal are not isolated, the networking mode is single, the device activation management is extensive, and the service mode is single, forms a complete closed loop of content collection, cloud analysis, result feedback and traffic operation, and can be extended and applied to various scenes such as homework correction, test paper evaluation and handwriting practice analysis.

Description

Technical Field Technical Field This invention relates to the fields of Internet of Things (IoT), artificial intelligence (AI), and cloud services, specifically to an intelligent evaluation system and its dual-track service method based on a dedicated content acquisition terminal and a cloud-based AI analysis engine. This invention is applicable to, but not limited to, student homework grading scenarios, and can be extended to various application scenarios requiring physical content acquisition and AI-based cloud-based analysis and evaluation, such as test paper evaluation, handwritten exercise analysis, drawing artwork evaluation, and knowledge test scoring. Background Technology

[0002] Currently, the digital collection and intelligent evaluation of various types of entity content face the following common problems: Limitations of using general-purpose devices for content collection. Using smartphones and other general-purpose devices for content collection (such as photographing assignments, exam papers, and documents) presents challenges including cumbersome procedures, susceptibility to interference from other applications on the device, and a lack of user-friendliness for children and the elderly. This is especially true in educational settings, where students are easily distracted by games and social media when using smartphones to photograph assignments.

[0003] Cloud-based AI analytics lacks synergy with dedicated hardware. Existing cloud-based AI analytics services (such as OCR recognition, content moderation, and knowledge assessment) typically receive input through general-purpose clients (web pages or mobile apps), lacking deep collaboration with dedicated data acquisition hardware. Dedicated hardware can optimize acquisition quality (angle, lighting, resolution), provide a one-click acquisition experience triggered by physical buttons, and enable device authentication and authorization management.

[0004] The drawback of a single service model. Existing similar products typically employ a uniform pricing model, requiring users to either purchase hardware and continue paying, or use the software for free but with limited functionality. The market lacks a flexible, dual-track service model that can simultaneously cover households that purchase dedicated hardware and users who only require software services.

[0005] Device activation and licensing management is rudimentary. Existing technologies for activating and managing dedicated hardware products typically employ simple serial number authentication, lacking a comprehensive mechanism for batch generation, batch tracking, automatic exclusion of anti-aliasing characters, one-time binding, and flexible licensing cycle management.

[0006] Dedicated data collection terminals rely on a single networking method. Existing dedicated IoT data collection terminals typically only support WiFi connectivity, making them unusable in scenarios with no WiFi coverage or unstable WiFi signals (such as school classrooms, outdoor learning corners, and elderly residences). Furthermore, terminals solely dependent on WiFi cannot leverage the wide-area coverage of cellular networks and lack cooperation channels with IoT communication operators in areas such as data traffic management and services, hindering the formation of a comprehensive business ecosystem that combines hardware sales with data traffic operations.

[0007] The application of cellular communication in IoT terminals is insufficient. Although 4G / 5G cellular communication modules have been widely used in IoT terminals, existing IoT products lack a complete solution that integrates cellular communication with dedicated content capture hardware and cloud-based AI analysis services. Particularly in educational scenarios, existing technologies have not yet offered effective solutions for how to incorporate cellular data costs into user subscription systems and how to negotiate data subsidies and revenue sharing with operators. Summary of the Invention

[0008] The purpose of this invention is to provide an intelligent evaluation system and its dual-track service method based on a dedicated content acquisition terminal and a cloud-based AI analysis engine. Through an innovative device code activation and dual-track service management mechanism, coupled with dedicated content acquisition hardware independent of user terminals and a multi-stage cloud-based AI analysis engine, it solves the problems of inefficient device activation management, a single service model, interference caused by the lack of physical isolation between the acquisition terminal and the user terminal, and the inability of dedicated terminals to adapt to wide-area coverage requirements due to their limited network connectivity. Simultaneously, this invention incorporates cellular data costs into the subscription system and supports revenue sharing with operators through a dual-mode communication design and an IoT traffic management mechanism, constructing a composite business ecosystem of hardware sales + software subscription + data traffic management. Technical solution

[0009] This invention provides an intelligent evaluation system based on a dedicated content acquisition terminal and a cloud-based AI analysis engine, comprising three main components: a content acquisition terminal, a cloud service platform, and a user terminal. The content acquisition terminal is an independent hardware device separate from the user terminal. Content acquisition and uploading can be completed with a single click, eliminating the need for complex shooting operations on the user terminal and achieving physical separation between the acquisition and viewing terminals. The content acquisition terminal supports dual-mode communication of WiFi and 4G / 5G cellular networks. When WiFi is available, it prioritizes using the WiFi channel to upload data to reduce data traffic costs. When WiFi is unavailable, it automatically switches to the cellular network to ensure timely data uploads and wide-area coverage.

[0010] Regarding device code management, this invention proposes a complete management mechanism that includes automatic exclusion rules for anti-confusion characters, batch generation, usage status tracking, and one-time binding of device codes, and realizes batch operation management through the management backend.

[0011] Regarding the dual-track service model, this invention innovatively divides users into two categories: device-activated users and non-device-code registered users, each subject to different subscription strategies. This differentiated service strategy balances both hardware sales and pure software service business models, significantly expanding the potential user base.

[0012] Regarding cloud-based AI analysis engines, this invention proposes a multi-stage pipeline architecture, comprising four stages: content preprocessing, content recognition, category determination, and item-by-item evaluation. This supports automatic determination and differentiated analysis across different subject areas or evaluation categories. When applied to homework grading scenarios, this engine is specifically implemented as an AI grading engine.

[0013] In the area of ​​IoT traffic management, this invention proposes an IoT traffic management module to monitor, manage quotas, connect with operators, and share revenue from cellular communication traffic. By incorporating cellular traffic costs into the user subscription system and supporting the conversion of operator traffic subsidies into subscription validity periods, a composite revenue model of "hardware sales + software subscription + traffic operation" is formed.

[0014] Device code anti-obfuscation mechanism: By excluding easily confused characters (such as I / O / 0 / 1), the error rate when users manually enter device codes is reduced, and the activation success rate is improved.

[0015] Batch management: Supports batch generation and tracking of device codes, facilitating inventory management and channel traceability in hardware production and sales.

[0016] The dual-track system offers flexible coverage: it serves both hardware purchasers and pure software users simultaneously, expanding the potential user base and lowering the barrier to entry for users with different willingness to pay.

[0017] Terminal physical isolation reduces interference: By physically separating the content acquisition terminal from the user terminal (smartphone), interference from mobile entertainment applications on the content acquisition process is isolated at the hardware level. The acquisition terminal retains only a single content acquisition function and does not include components that may distract attention, such as browsers and app stores.

[0018] Automated network configuration: Enables network configuration without application installation through AP hotspot + Web configuration page technology, reducing the barrier to entry.

[0019] Multi-scenario adaptability: Through a configurable cloud-based analytics engine, the system can expand from homework grading to intelligent assessment of various knowledge-based content such as test paper evaluation, handwritten exercise analysis, and drawing evaluation.

[0020] Storage cost optimization: By using tiered storage and automatic expiration strategies, cloud storage costs can be effectively controlled while ensuring service quality.

[0021] Historical user compatibility: The compatibility handling module ensures that the rights and interests of early lifetime users are not affected after the system is upgraded to a dual-track system.

[0022] Dual-mode communication for wide-area coverage: With a dual-mode communication design of WiFi + cellular, the terminal can upload data at zero cost when WiFi is available and automatically switch to the cellular network when WiFi is unavailable, achieving wide-area coverage. It is suitable for various deployment environments such as schools, outdoors, and mobile scenarios.

[0023] Cellular data traffic is incorporated into the subscription system: Through the IoT data traffic management module, the data traffic costs generated by cellular communication are incorporated into the user subscription fee system, and operator data traffic subsidies are converted into subscription validity period, forming a composite revenue model of "hardware sales + software subscription + data traffic operation".

[0024] Operator cooperation and traffic management: By connecting with the API of IoT communication operators, it enables bulk purchase of traffic packages, allocation by terminal and automatic reconciliation and settlement, supports traffic subsidies and revenue sharing mechanisms, and creates additional traffic management revenue for platform operators.

[0025] Instruction Manual (Part Three: Illustrations) Attached Figure Description

[0026] Figure 1 is a schematic diagram of the overall architecture of the system of the present invention; Figure 2 is a hardware structure block diagram of the image acquisition terminal; Figure 3 is a schematic diagram of the device code generation and management process; Figure 4 is a schematic diagram of the dual-track user service management process; Figure 5 is a schematic diagram of the network configuration process for the image acquisition terminal; Figure 6 is a schematic diagram of the process of uploading assignment images and AI grading; Figure 7 is a schematic diagram of the lifecycle management of job image storage; Figure 8 is a schematic diagram of the dual-mode communication network selection and switching process; Figure 9 is a schematic diagram of the IoT traffic management and operator integration process.

[0027] Instruction Manual (Part Four: Detailed Implementation) Detailed Implementation

[0028] The present invention will now be described in further detail with reference to the accompanying drawings.

[0029] Example 1: System Overall Architecture As shown in Figure 1, this system consists of three parts: an image acquisition terminal 100, a cloud service platform 200, and a user terminal 300. The image acquisition terminal 100 is a dedicated hardware device that connects to the internet via Wi-Fi or other wireless methods and communicates with the cloud service platform 200. The cloud service platform 200 is deployed on a cloud server and is responsible for device code management, user account management, AI-based grading, and notification push notifications. The user terminal 300 is a software client implemented as a WeChat mini-program (or other lightweight application) that runs on the parent's smartphone.

[0030] Example 2: Hardware Structure of Image Acquisition Terminal As shown in Figure 2, the image acquisition terminal 100 includes the following hardware modules: Wireless communication module 101: supports dual-mode communication, including a WiFi communication submodule 101a and a cellular communication submodule 101b; WiFi communication submodule 101a: can use the WiFi module built into the ESP32 series chip or other equivalent IoT communication chips, supporting dynamic switching between STA (site) mode and AP (access point) mode; Cellular communication submodule 101b: can use a 4G Cat.1 / Cat.4 module (such as EC800, SIM7600, etc.) or a 5G NR module, connecting to an IoT SIM card provided by a communication operator via SIM card interface 101c, or remotely configuring the operator's configuration file to activate cellular communication services via eSIM chip 101d; the cellular communication submodule supports uploading data to the cloud service platform 200 via 4G / 5G cellular networks when there is no WiFi coverage; Image acquisition module 102: can use OV2640 or other equivalent CMOS. The image sensor supports at least UXGA (1600×1200) resolution and outputs images in common compression formats such as JPEG; the processor module 103 can use an ESP32 series dual-core processor (240MHz clock speed, built-in SRAM and PSRAM), or other equivalent microcontrollers, to run embedded firmware to control the photo taking and uploading process; the processor module 103 has a built-in network selection strategy, which prioritizes the detection of WiFi availability during data upload, and automatically switches to the cellular communication submodule 101b when WiFi is unavailable; the trigger module 104 includes at least one physical button, which supports distinguishing between short press and long press operations; the status indicator module 105 includes at least one LED indicator or equivalent visual indicator element, which provides feedback on the device status through different flashing modes; the non-volatile storage module 106 can use the processor's built-in NVS (Non-Volatile Storage) Flash partition or an external EEPROM / Flash chip to persistently store the device identification code, WiFi configuration information, cellular network configuration parameters, and system configuration parameters. When the image acquisition terminal 100 leaves the factory, it obtains a unique device code from the cloud service platform 200 and writes it into the protected partition in the non-volatile storage module 106 through a burning tool. The data in this partition cannot be modified or erased by normal user operations during the device's life cycle.

[0031] Example 3: Device Code Generation and Management Mechanism As shown in Figure 3, the device code generation and management process of this invention is detailed below: Device Code Format Definition: The device code consists of 8 characters, with the character set being uppercase English letters (AZ) and numbers (0-9), excluding the uppercase letters "I" and "O" which are easily confused with numbers. There are a total of 34 valid characters (24 letters + 10 numbers, excluding 2), theoretically generating approximately 1.78 trillion unique device codes. Batch Generation: The management backend (Web management interface) of the cloud service platform 200 provides a batch generation function for device codes. After the administrator inputs the number to be generated, the system generates device codes in batches according to preset rules. Each device code is associated with a unique batch number (e.g., the format is "BATCH-YYYYMMDD-NNN"). The data structure of the device code record includes: device code string, batch number, generation time, usage status (enumeration value: unused / used), activated user identifier (default is empty), and activation time (default is empty). Batch Management Function: The management backend supports filtering and viewing by batch number, paginated browsing, and exporting CSV files by batch. Each batch can be independently exported as a list of device codes for use in production and shipping. Anti-duplicate and security measures: Each device code is marked as unique in the database, and an activated device code cannot be bound again; the device code verification interface has a frequency limit to prevent brute-force attacks.

[0032] Example 4: Dual-track user service management mechanism As shown in Figure 4, the dual-track user service management process of this invention is detailed below: User type determination: When a user registers on the user terminal 300, the system automatically determines the user type based on whether the registration request contains a valid device code: if the request contains a device code and the cloud service platform 200 verifies it, the user is classified as a device user; if the request does not contain a device code or is registered using a mobile phone number, the user is classified as a mobile phone registered user. Device user service strategy (first track): After the user enters a valid device code, the cloud service platform 200 binds the device code to the user account, and the device code status is updated to activated; the user account's subscription status is automatically set to valid, and the expiration date is the binding date plus 365 days (first year free strategy); during the subscription validity period, the user can use the AI ​​grading function without restriction to view the complete grading results; before the expiration date, the system pushes a renewal reminder through the user terminal 300 and displays the annual renewal entry (e.g., ¥99 / year); the device code binding is unique: one device code can only be bound to one user account, and it cannot be unbound or transferred after binding. Mobile Registration User Service Strategy (Second Track): Registered users automatically receive a 30-day full-featured trial period; during the trial period, the user experience is the same as for device users; after the trial period, the system displays renewal options on the user's terminal 300: monthly payment (e.g., ¥9.9 / month) or annual payment (e.g., ¥99 / year); after payment, the subscription status is updated to valid; users who have not renewed can still view their history but cannot submit new assignments or obtain grading results. Historical User Compatibility: For early lifetime valid subscription users existing before the system upgrade, the cloud service platform 200's compatibility processing module automatically maps them to a valid status in the subscription query interface and sets a future expiration date (e.g., December 31, 2099) to ensure their original rights are not affected. The front end displays the corresponding renewal entry based on the user type field.

[0033] Example 5: Multi-stage pipeline for AI grading engine As shown in Figure 6, the workflow of the AI ​​grading engine module is detailed below: First Stage – Image Preprocessing: After receiving the original assignment image, the following preprocessing steps are performed: Orientation Detection and Automatic Correction: Detecting the tilt angle of text lines and performing rotation correction; Perspective Transformation: Correcting perspective distortion caused by the shooting angle and restoring the rectangular plane of the paper; Image Enhancement: Including adaptive contrast adjustment, noise reduction, and sharpening to improve the accuracy of subsequent OCR recognition. Second Stage – Optical Character Recognition: The preprocessed image is fed into the OCR recognition model to extract text regions and text content from the image. The OCR model can be based on mature deep learning architectures (such as CRNN, Transformer-based OCR, etc.) and supports mixed Chinese and English recognition. The recognition results are organized by text line or region, preserving text position information. Third Stage – Automatic Subject Classification: Semantic analysis is performed on the extracted text content, and the text classification model automatically determines the subject category of the assignment, such as mathematics, Chinese, English, etc. The classification criteria include keyword matching, formula symbol density analysis, and contextual semantic understanding. The subject classification result determines which subject grading logic unit is subsequently called. The fourth stage – Question-by-question grading and result generation: Based on the subject category, the corresponding grading logic unit is invoked: Mathematics grading unit: analyzes arithmetic expressions, equations, and geometry problems, verifying the correctness of answers through a symbolic computation engine; Chinese grading unit: analyzes the standardization of character writing, the correctness of word collocation, and the grammatical structure of sentences; English grading unit: checks word spelling, grammatical structure, and sentence pattern usage. After grading is completed, the result generation submodule outputs structured data, including: subject category, question-by-question scoring details (question number, question content extraction, student's answer, correctness judgment, and feedback text for incorrect questions), total score (accuracy percentage, score, and grade), and personalized learning suggestions (such as hints for weak knowledge points and recommendations for similar practice questions).

[0034] Example 6: Network Configuration Process for Image Acquisition Terminal As shown in Figure 5, the network configuration process of the image acquisition terminal 100 is as follows: Step 1: The terminal powers on and starts up. The processor module 103 checks whether there is a valid WiFi SSID and password in the non-volatile storage module 106. Step 2: If there is no WiFi configuration, the processor module 103 controls the wireless communication module 101 to switch to AP mode. In AP mode, a WiFi hotspot with an SSID containing the product identification prefix (such as "Parent AI Assistant - Network Configuration") is created. At the same time, the built-in DNS server and HTTP Web server are started, and the DNS request is hijacked to the local IP address (such as 192.168.4.1) to support the automatic pop-up function of the Captive Portal. Step 3: After the user connects to the WiFi hotspot using a smartphone or other terminal device, the operating system automatically detects the Captive Portal and pops up the network configuration page (or the user manually accesses any URL through a browser and is redirected to the network configuration page). Step 4: The network configuration page displays the following information: device number, and a list of available WiFi networks in the vicinity scanned by the image acquisition terminal through the wireless communication module 101 (including SSID, RSSI signal strength value, and security encryption type). Step 5: The user selects the target WiFi network and enters the password on the page before submitting. After receiving the configuration, the firmware switches the wireless communication module 101 to STA mode and attempts to connect to the target WiFi. Step 6: Connection successful: The WiFi SSID and password are saved to the non-volatile storage module 106, and the LED indicator displays a success message (solid on for 2 seconds). After 3 seconds, it automatically restarts and enters normal working mode. Connection failed: The wireless communication module 101 reverts to AP mode, and the network configuration page displays the specific error reason (such as incorrect password, weak signal, etc.), allowing the user to reconfigure.

[0035] Example 7: Job Image Upload and Retry Mechanism The image upload process of the image acquisition terminal 100 includes the following reliability assurance mechanisms: After taking a picture, the processor module 103 packages the JPEG image data (approximately 80-150KB, resolution 1600×1200) and the device terminal identifier, and uploads it to the designated API endpoint of the cloud service platform 200 via HTTP POST multipart / form-data; the upload interface returns a task ID as a confirmation of receipt; after the first upload fails (network timeout or server error), the processor module 103 automatically waits for 2 seconds and then retryes, up to 3 times; if all 3 retries fail, the LED indicator outputs a failure indication mode (rapid flashing at 100ms intervals), and returns to the ready state after 3 seconds; after the network connection is restored, the processor module 103 automatically resubmits the unsuccessfully uploaded images temporarily stored in PSRAM for upload (cleared if power is lost, data can be retaken); when the upload is successful, the LED indicator stays on for 2 seconds and then turns off, indicating that the task has been submitted to the server processing queue.

[0036] Example 8: Storage Lifecycle Management As shown in Figure 7, the cloud service platform 200 implements the following tiered storage management for homework image data: Tier 1 (Hot Storage): After uploading, homework images are saved at the standard access level, allowing parents to view them in high definition at any time, with a retention period of 30 days; Tier 2 (Warm Storage): After 30 days, the images are automatically migrated to the low-frequency access level (storage costs are reduced by approximately 50%), where parents can still access them but the loading speed is slightly slower; Tier 3 (Cold Storage / Deletion): After 365 days, a notification message is pushed to the user's terminal 300 before the original images are automatically deleted. Users can choose to download and save the original images for free or pay to extend the storage period; Data Processing Retention: After the original image files are deleted, the structured correction results data (plain text, approximately 1-2KB per file) is retained long-term, allowing parents to view historical correction records at any time. This tiered management solution can reduce cloud storage costs by more than 60% compared to full standard storage, given the expected user scale.

[0037] Example 9: Management Backend and Multi-Role Permissions The cloud service platform 200 provides a web-based management backend that supports the following management functions: Super administrator accounts can create, modify, and delete ordinary administrator accounts, enabling multi-role hierarchical management; the device code batch generation interface supports inputting the generation quantity, automatically generating batch numbers, exporting CSV files, and tracking usage by batch; ordinary administrators can view the device code list, export device codes, and generate new device codes, but cannot access user privacy data; all management operations are recorded with timestamps and the user performing the operation, supporting operation auditing.

[0038] Example 10: LED Status Indicator Comparison LED Status Indicator Table: Rapid flashing (250ms on / 250ms off): Network pairing mode, waiting for network pairing; Medium flashing (500ms on / 500ms off): Connecting to WiFi; Slow flashing (100ms on / 2900ms off): Ready state, waiting to take a picture; Rapid flashing (50ms on / 50ms off): Taking a picture; Rapid alternation (150ms on / 150ms off): Uploading data; Constant light for about 2 seconds: Upload successful; Rapid flashing (100ms on / 100ms off): Upload failed; One-second alternation (1000ms on / 1000ms off): Resetting factory settings.

[0039] Example 11: Restore factory settings After the user presses and holds the reset button in the trigger module 104 for a preset duration (e.g., 8 seconds), the processor module 103 executes the factory reset process: clearing the WiFi SSID and password in the non-volatile storage module 106; retaining the device code and terminal identifier (data in the protected partition that cannot be erased); the LED indicator enters a one-second alternating flashing mode for 3 seconds; the device automatically restarts and enters the network configuration mode. This design ensures that when the device needs to be transferred to other users, the WiFi configuration of the previous user's home can be cleared, while maintaining the immutability of the device's identity.

[0040] Example 12: Dual-mode communication network selection and switching As shown in Figure 8, the content acquisition terminal 100 of the present invention supports dual-mode communication of WiFi and cellular networks. The processor module 103 executes the following network selection and switching strategy: Network detection and selection: When the terminal starts or prepares to upload data, the processor module 103 first detects whether the WiFi communication submodule 101a has been connected to an available WiFi network; if WiFi is available and the signal strength is higher than a preset threshold, data is uploaded through the WiFi channel (zero data cost); if WiFi is unavailable or the signal is lower than the threshold, the cellular communication submodule 101b is automatically activated to upload data through the 4G / 5G cellular network configured by the SIM card or eSIM; if both WiFi and cellular networks are unavailable, the terminal temporarily stores the image in PSRAM, enters a waiting state, and periodically retryes network detection. Cellular Communication Activation and Configuration: Supports physical SIM card hot-swapping detection: When a SIM card is detected, the system automatically reads the SIM card's ICCID and IMSI information and initializes the cellular network registration process; Supports eSIM remote configuration: After the terminal connects to the cloud service platform 200 via WiFi, it can receive the eSIM configuration file issued by the operator and remotely activate the cellular communication service without a physical SIM card; Cellular communication parameters (APN, authentication information, etc.) are stored in the non-volatile storage module 106 and are automatically restored after a restart. Transparency of Communication Channel Switching: Regardless of whether the data is uploaded via WiFi or cellular network, the receiving interface and processing flow of the cloud service platform 200 are completely identical; The uploaded data includes the terminal identifier and communication channel identifier, which the cloud service platform 200 uses to statistically analyze the traffic usage of each terminal; For users, the network switching process is completely transparent, and users only need to operate the trigger module to complete the data collection and uploading.

[0041] Example 13: IoT Traffic Management and Operator Cooperation Mechanism As shown in Figure 9, the IoT traffic management module of the cloud service platform 200 implements the following functions: Traffic monitoring and quota management: The traffic monitoring submodule provides real-time statistics on the uplink data traffic generated by each content collection terminal through the cellular network, recording traffic usage details by terminal identifier (device code) and date; the management backend provides a traffic usage dashboard, allowing administrators to view traffic usage statistics for individual terminals or batches; the traffic quota management submodule sets monthly traffic quotas for each terminal (e.g., default 500MB / month), and pushes reminder notifications to user terminals when traffic usage reaches 80% and 100% of the quota, respectively. Carrier integration and settlement: The carrier integration submodule connects with the traffic management platform of IoT communication operators via API to achieve: Batch purchase of traffic packages: Administrators can purchase traffic packages for a batch of terminals in the management backend; Traffic allocation by terminal: The purchased traffic pool is allocated to each terminal according to the terminal dimension; Traffic usage data reconciliation: Traffic usage details are periodically retrieved from the operator platform and reconciled with local statistical data for verification. Traffic subsidies and revenue sharing: When a partner operator provides traffic subsidies for a terminal, the traffic billing submodule will convert the subsidy amount into a proportional extension of the subscription validity period for the corresponding user account (e.g., an operator subsidizing ¥10 of traffic is equivalent to approximately one month of renewal). When an operator incorporates the cellular traffic generated by the terminal into its IoT traffic management system and pays a revenue share to the platform operator, the traffic billing submodule will calculate the traffic management revenue that the platform operator should receive according to the agreed revenue sharing ratio. The traffic management revenue can be distributed among the platform operator, equipment manufacturer, and traffic provider according to the proportion agreed in the contract, forming a composite revenue model of hardware sales + software subscription + traffic operation.

Claims

1. An intelligent evaluation system based on a dedicated content acquisition terminal and a cloud-based analysis engine, characterized in that, include: Content acquisition terminals, cloud service platforms, and user terminals; The content acquisition terminal is an independent hardware device separate from the user terminal, comprising: a wireless communication module for establishing a data connection with the cloud service platform via a wireless network, the wireless communication module including at least one of a WiFi communication submodule and a cellular communication submodule; the cellular communication submodule supports the insertion of a user identity card (SIM / eSIM) and establishes a data connection with the cloud service platform via a 4G or 5G cellular network; a content acquisition module for acquiring optical images of the entity content to be evaluated, the entity content including but not limited to: handwritten assignments, printed test papers, exercise book pages, drawings, handwritten exercises, and written works; a processor module electrically connected to the wireless communication module and the content acquisition module respectively, for controlling the content acquisition process and the data upload process; a trigger module electrically connected to the processor module for receiving user operation commands to initiate a single content acquisition; a status indication module electrically connected to the processor module for indicating the terminal's working status through visual signals; and a non-volatile storage module electrically connected to the processor module for persistently storing terminal identification information and system operating parameters; the cloud service platform includes: a device code. The system comprises the following modules: a management module for batch generating device activation codes and tracking the lifecycle of each code; a user account management module for managing user accounts and differentiating user types based on whether a user activated their device using a device code; a subscription and permission management module for implementing differentiated subscription strategies for different user types; a cloud analytics engine module for performing multi-stage automated analysis and evaluation of uploaded content images based on artificial intelligence, generating structured evaluation results; and a notification module for pushing evaluation results to user terminals. The user terminals include: a device activation unit for receiving device codes input by the user and binding the device codes to the user account; a subscription management unit for displaying the corresponding subscription status and renewal entry based on user type; and a result viewing unit for displaying evaluation results and historical records.

2. The system according to claim 1, characterized in that, The cloud-based analysis engine module includes: a content preprocessing submodule for performing orientation correction, perspective transformation, and image enhancement on the original image; a content recognition submodule for extracting text, symbols, and graphic content from the preprocessed image; a category determination submodule for automatically determining the subject area or assessment category to which the extracted content belongs; a set of assessment execution submodules, containing assessment logic units for different subject areas or assessment categories, each of which performs item-by-item judgment on the extracted content according to the knowledge base and assessment rules of the corresponding field; and a result generation submodule for integrating the assessment results into structured data output, wherein the structured data includes at least: assessment category, item-by-item scoring details (item identifier, question content, answer content, correctness judgment, feedback information), total score, and analysis suggestions.

3. The system according to claim 1, characterized in that, The device code generation rules of the device code management module include: the device code consists of a preset length of alphanumeric combination, excluding characters that are easily confused in appearance, and including at least two of the uppercase letter "I", uppercase letter "O", number "0", and number "1"; device codes are generated in batches in the management backend of the cloud service platform, and each batch is associated with a unique batch number; each device code record contains the following fields: device code string, batch number, generation time, usage status (not activated / activated), activated user identifier, and activation time; the management backend supports filtering by batch number, paginated viewing, and exporting the device code list by batch; activated device codes cannot be used again to bind other user accounts.

4. The system according to claim 1, characterized in that, The dual-track strategy of the subscription and permission management module specifically includes: Track 1 – Device Users: After a user enters a valid device code and completes binding through the device activation unit, the user's account subscription status is automatically set to activated, with a validity period of a preset number of days (e.g., 365 days) starting from the binding date. An annual renewal option is provided after the validity period expires. Track 2 – Mobile Phone Registered Users: When a user registers with a mobile phone number without binding a device code, the user account receives a preset trial period (e.g., 30 days). A monthly or annual renewal option is provided after the trial period expires. During the subscription validity period, users can use the revise and view functions normally. After the validity period expires, the system restricts the complete viewing of revise results but retains historical access records.

5. The system according to claim 1 or 2, characterized in that, When applied to homework grading scenarios: the content acquisition terminal is specifically implemented as a dedicated homework photo-taking terminal; the entity content is the student homework to be graded; the cloud analysis engine module is specifically implemented as an AI grading engine module, used to perform multi-stage automatic recognition and question-by-question grading of the uploaded homework images; the evaluation result is specifically a structured grading result, including question-by-question scoring details, total score, and learning suggestions.

6. The system according to claim 1, characterized in that, The network configuration process of the content acquisition terminal includes: after the content acquisition terminal starts up, the processor module detects whether there is valid network configuration information in the non-volatile storage module; if not, the content acquisition terminal switches to wireless access point mode and creates a local wireless hotspot, the identifier of which includes a product identification prefix; after the user connects to the wireless hotspot through a terminal device, they access the local network configuration page through a browser, the configuration page displays a list of available wireless networks scanned by the content acquisition terminal and their corresponding signal strength indicators; after the user selects a target wireless network and enters access credentials on the configuration page, the content acquisition terminal attempts to connect to the target wireless network in site mode; after a successful connection, the network configuration information is saved to the non-volatile storage module and the terminal automatically restarts to enter normal working mode; after a connection failure, the terminal automatically reverts to wireless access point mode, allowing the user to reconfigure.

7. The system according to claim 1, characterized in that, The triggering module supports a first operating mode and a second operating mode: the first operating mode corresponds to a trigger signal of a first duration, which triggers the content acquisition module to acquire content and upload it to the cloud service platform via the wireless communication module using the HTTP protocol; the second operating mode corresponds to a continuous trigger signal of a second duration, wherein after the second duration is longer than the first duration and reaches a preset threshold, the network configuration information in the non-volatile storage module is cleared, so that the content acquisition terminal is restored to the factory network state.

8. The system according to claim 1, characterized in that, The status indicator module distinguishes at least six states by controlling different flashing sequences of the light-emitting element: network distribution waiting state, network connection in progress state, ready to collect data state, data collection in progress state, data uploading in progress state, upload successful state, upload failed state, and factory reset state. Each state corresponds to a unique combination of flashing period and duty cycle.

9. The system according to claim 1, characterized in that, The wireless communication module includes a WiFi communication submodule and a cellular communication submodule. The content acquisition terminal supports dual-mode communication. When uploading data, the processor module executes a network selection strategy: it first checks whether the WiFi communication submodule is connected to an available WiFi network; if WiFi is available, it uploads data via the WiFi channel; if WiFi is unavailable, it automatically switches to the cellular communication submodule to upload data via a 4G or 5G cellular network. The cellular communication submodule includes a SIM card interface or an eSIM chip, supporting the insertion of a SIM card from a telecommunications operator or activation of cellular communication services by remotely configuring an eSIM configuration file. When both WiFi and cellular networks are available simultaneously, the content acquisition terminal selects the communication channel according to a preset strategy, which may prioritize based on data cost, transmission speed, or signal strength.

10. The system according to claim 9, characterized in that, The cloud service platform includes an IoT traffic management module for managing the data traffic of the cellular communication submodule and operator cooperation; and a traffic monitoring submodule for real-time monitoring of the data traffic generated by each content collection terminal through the cellular network and recording traffic usage details according to terminal identification. The traffic quota management submodule is used to set cellular data traffic quotas for each content collection terminal and send a reminder notification to the user terminal when the traffic usage reaches the quota threshold. The operator interface submodule is used to interface with the traffic management platform of IoT communication operators to realize bulk purchase of traffic packages, allocation by terminal, and reconciliation of traffic usage data. The traffic billing submodule is used to calculate the traffic cost for each terminal based on traffic usage details and operator settlement rates. It supports including traffic costs in the device user's annual renewal fee or having them subsidized by the operator.

11. The system according to claim 10, characterized in that, The IoT traffic management module further supports traffic subsidies and revenue sharing mechanisms: when the content collection terminal uses IoT traffic packages provided by the cooperating operator, the cloud service platform records traffic usage data and conducts regular reconciliation and settlement with the operator; the traffic billing submodule supports converting the traffic subsidy amount provided by the operator into an extension of the user account's subscription validity period, or distributing the revenue sharing generated from traffic usage between the platform operator and the traffic provider proportionally.

12. The system according to claim 1, characterized in that, The cloud service platform implements the following lifecycle strategy for content image data: First storage stage: After the content image is uploaded, it is saved at the standard storage level; Second storage stage: After the first preset time period has elapsed, the content images will be migrated from the standard storage level to the low-frequency or archive storage level; Third storage stage: After the second preset time period, a reminder notification is sent to the user terminal before the original image file is automatically deleted, prompting the user to choose to download and save the original image; wherein, after the original image file is deleted in the third storage stage, the corresponding structured evaluation result data is retained.

13. The system according to claim 1, characterized in that, Further includes: The compatibility processing module is used to identify early permanently valid subscribers in the system, convert the early permanently valid subscribers into the same activated state as the device users in the dual-track strategy in the query interface, and set a future expiration date to ensure their continued validity.

14. A method for activating and evaluating content of a content acquisition terminal, applied to the system according to any one of claims 1 to 13, characterized in that, The process includes the following steps: Step A1: Deploy the content acquisition terminal independently of the user terminal in the content acquisition scenario. The content acquisition terminal has a unique device code written to it at the factory. Step A2: The user enters the device code in the user terminal to initiate a device activation and binding request to the cloud service platform. Step A3: The cloud service platform verifies the validity and usage status of the device code. After successful verification, the device code is bound to the user account and the usage status is marked as activated. Step A4: After successful activation, the cloud service platform automatically grants the corresponding subscription permissions according to the user type: if the user is a device code activation user, a valid subscription period of a preset number of days is automatically granted. If the user is not a device code registered user, a trial period of a preset number of days is granted; Step A5: The user operates the content acquisition terminal to collect content, and the content acquisition terminal uploads the collected content image and terminal identifier to the cloud service platform via wireless network; Step A6: After receiving the content image, the cloud service platform performs multi-stage automatic analysis based on artificial intelligence through the cloud analysis engine to generate structured evaluation results; Step A7: The cloud service platform associates the evaluation results with the user account and pushes them to the user terminal through the notification module. The user terminal displays the corresponding evaluation result content according to the user's subscription permissions.

15. The method according to claim 14, characterized in that, The multi-stage processing flow of the cloud-based analysis engine includes: First stage: orientation correction, perspective transformation, and image enhancement of the original content image; Second stage: extraction of text and symbol content from the preprocessed image through optical character recognition; Third stage: automatic determination of the subject area or evaluation category to which the entity content belongs based on the extracted content; Fourth stage: invoking the evaluation logic unit corresponding to the determined category, combining the domain knowledge base to analyze and determine the content item by item, and generating structured evaluation results.

16. The method according to claim 14, characterized in that, The generation and management of the device codes include: the management backend of the cloud service platform generates device codes in batches according to preset character set rules, wherein the character set excludes characters that are easily confused in appearance, including at least two of the uppercase letter "I", uppercase letter "O", number "0" and number "1"; each device code is associated with a unique batch number, and the generation time and initial usage status are recorded; the management backend supports filtering by batch number, paginated viewing, and batch export of device code lists; the device code verification interface is set with a frequency limit to prevent brute-force enumeration attacks.

17. The method according to claim 14, characterized in that, The differentiated subscription permission management described in step A4 further includes: after the expiration of the validity period of a user activated by a device code, a renewal reminder is pushed to the user's terminal and an entry for periodic renewal is displayed; after the trial period of a user registered without a device code expires, a periodic payment option is displayed on the user's terminal, and the subscription status is updated to valid after the user completes the payment; for historically existing permanently valid subscription users, the query interface is compatiblely mapped to a valid status and a future expiration date is set.

18. The method according to claim 14, characterized in that, The content acquisition terminal's upload process includes a failure retry mechanism: it automatically retryes after an upload failure, with the number of retries not exceeding a preset limit; If the retries still fail after reaching the limit, a failure indication will be output through the status indication module, and the incomplete data in the queue will be automatically re-uploaded after the network connection is restored.

19. The method according to claim 14, characterized in that, The content acquisition terminal uploads data via dual-mode communication. The process includes the following steps: Step B1: When the content acquisition terminal triggers data upload, it first checks whether the WiFi communication submodule is connected to an available WiFi network; Step B2: If WiFi is available and the signal strength is higher than a preset threshold, data is uploaded through the WiFi channel; Step B3: If WiFi is unavailable or the signal is lower than the threshold, the processor module automatically activates the cellular communication submodule and uploads data through the 4G / 5G cellular network configured by the SIM card or eSIM; Step B4: The uploaded data contains the terminal identifier and the communication channel identifier, which the cloud service platform uses to record the terminal's traffic usage; Step B5: If both WiFi and the cellular network are unavailable, the terminal temporarily stores the collected content in memory and periodically retryes network detection until the network is restored and then automatically uploads the data.

20. The method according to claim 19, characterized in that, The cloud service platform performs the following management steps for cellular traffic: Step C1: Monitor the data traffic generated by each content collection terminal through the cellular network in real time, and record traffic usage details by terminal identifier and date; Step C2: Set cellular data traffic quotas for each terminal, and send reminder notifications to user terminals when traffic usage reaches the quota threshold; Step C3: Connect with the traffic management platform of IoT communication operators through API to realize bulk purchase of data packages, allocation by terminal, and reconciliation of traffic usage data; Step C4: Calculate the traffic cost of each terminal based on traffic usage details and operator settlement rates, convert the traffic subsidies provided by operators into an extension of the user account subscription validity period, or distribute the traffic operation revenue proportionally between the platform operator and the traffic provider.

Images