Intelligent picture quality detection system and method

Abstract

The invention provides an intelligent picture quality detection system and method, and the system comprises a picture data input and synchronization module which is used for importing and updating original picture data; the automatic detection module is used for detecting the original picture data and dividing the original picture data into qualified pictures and unqualified pictures; the unqualified picture recognition module is used for processing unqualified pictures and reasons thereof; the picture optimization module is used for optimizing the quality of the unqualified pictures and transmitting the unqualified pictures to the qualified picture storage module; the qualified picture storage module is used for storing the optimized pictures and summarizing the qualified pictures into a qualified data set; and the machine learning module is used for training and optimizing the model by using a qualified data set and detecting, identifying and classifying original pictures. According to the method, the platform and the enterprise can be helped to quickly identify problems and optimization points in the picture, specific improvement suggestions are provided, the product display effect and the customer purchase experience are improved, and the picture quality detection efficiency and accuracy are greatly improved, so that the enterprise is helped to optimize a product display strategy.

Description

Technical Field

[0001] The present invention relates to the fields of image processing and computer vision, and in particular to an intelligent image quality detection system and method, and more particularly to a system and method for detecting the quality of industrial product images. Background Art

[0002] With the increasing popularity of e-commerce procurement, online display and sales of industrial products have become a crucial channel for manufacturers and buyers. High-quality product images play a crucial role on e-commerce platforms, not only influencing customers' initial impressions but also directly influencing purchasing decisions and a company's brand image. High-quality images clearly showcase product details and features, thereby attracting more potential customers.

[0003] However, traditional image quality inspection methods rely primarily on manual inspection. These methods are not only time-consuming and labor-intensive, but also susceptible to human error, leading to errors and inconsistencies in inspection results. Furthermore, as the variety and number of products on the platform increase, the efficiency and accuracy of manual inspections are unable to meet the growing demand.

[0004] Patent document CN111340749B discloses an image quality detection method, device, equipment and storage medium, which aims to solve the problem that the detection results of image brightness in traditional technology are unstable, the detection form of clarity is relatively single, and it is unable to detect images with different levels of problem richness. However, the technical method used is essentially different from the present invention.

[0005] Therefore, there is an urgent need for an intelligent method and system that can efficiently detect and evaluate the quality of industrial product images on e-commerce platforms. Summary of the Invention

[0006] In view of the defects in the prior art, the object of the present invention is to provide an intelligent image quality detection system and method.

[0007] According to the present invention, an intelligent image quality detection system is provided, comprising:

[0008] Image data entry and synchronization module, used to import and update original image data;

[0009] Original picture storage module, used to store original picture data;

[0010] An automatic detection module is used to detect the original image data and classify it into qualified images and unqualified images;

[0011] Unqualified image recognition and cause analysis module, used to analyze and process the reasons for unqualified images;

[0012] The image optimization module is used to optimize the image quality of unqualified images and transfer the optimized images to the qualified image storage module;

[0013] The qualified image storage module is used to store optimized images and qualified images, and organize and summarize them into qualified data sets;

[0014] The machine learning processing result module uses qualified data sets to continuously train and optimize the model, and uses the optimized model to detect, identify and classify the original images.

[0015] Preferably, it also includes:

[0016] The user interaction module is used to display detection results, quality scores, recognition classifications, and optimization suggestions; users can manually correct and adjust the image based on the recognition results to complete the interaction with the system.

[0017] Preferably, the picture data entry and synchronization module can automatically import picture data in batches from different supplier systems, update them synchronously, and use a data transmission protocol to maintain picture information.

[0018] Preferably, the original image storage module includes the use of redundant storage technology and provides basic data support for image quality detection.

[0019] Preferably, the automatic detection module performs image quality detection based on a deep learning model and automatically identifies image quality problems; the image quality problems include image blur problems, image size problems, image proportion problems, and image background color problems.

[0020] Preferably, the unqualified image recognition and cause analysis module can classify the problems of unqualified images and generate an unqualified analysis report.

[0021] Preferably, the machine learning processing result module continuously analyzes the detection and optimization results through a machine learning algorithm to optimize the parameters of the learning model; and automatically updates the learning model through a feedback mechanism.

[0022] According to the present invention, an intelligent image quality detection method is provided, comprising:

[0023] Step S1: Collect synchronized product images and their corresponding product information and classification data from different suppliers;

[0024] Step S2: pre-processing the collected product images and their corresponding product information and classification data to obtain pre-processed results; the pre-processing includes denoising, standardization, size adjustment, and format conversion of the product images;

[0025] Step S3: Using a deep learning model to automatically detect and identify the category of the product after preprocessing the product image to obtain the recognition result;

[0026] Step S4: Classify the product images into qualified images and unqualified images according to the recognition results;

[0027] Step S5: Use the deep learning model to make improvement suggestions for unqualified images;

[0028] Step S6: Manually check the unqualified images and make further adjustments to the products and categories;

[0029] Step S7: Storing qualified images and adjusted and corrected images so that the system can continuously train and optimize the machine learning model based on these images;

[0030] Step S8: Use the optimized machine learning model to identify and detect product images.

[0031] Preferably, step S3 includes the following sub-steps:

[0032] Step S3.1: Use a deep learning model to extract product image features and perform quality inspection;

[0033] Step S3.2: Generate a quality score for each product image and mark specific quality issues.

[0034] Preferably, an automatic learning mechanism is used to extract key features from product images and associate them with product classification information; transfer learning and adaptive learning techniques are used to adapt the model to the recognition needs of new products and new classifications.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1. The system provided by the present invention adopts deep learning technology and automated processing flow, which can identify and evaluate the quality of industrial product images provided by manufacturers. By accurately identifying the quality of images, the system can quickly provide improvement suggestions to users and point out any problems in detail. This function not only helps users improve the display effect of images, but also improves the customer's product selection experience.

[0037] 2. The system provided by the present invention can not only help platforms and enterprises quickly identify problems and optimization points in pictures, but also provide specific improvement suggestions to enhance the product display effect and customer purchasing experience; by applying this system, the platform will greatly improve the efficiency and accuracy of image quality detection, thereby helping enterprises optimize product display strategies and enhance the effectiveness of their online marketing, and has good practicality.

[0038] 3. The present invention can perform quality inspection on image data provided by different suppliers, and supports batch addition, synchronization, inspection, unqualified quantity statistics, and image editing and optimization; the system uses qualified commodity and classification data to continuously train and optimize the machine learning model, continuously improving the accuracy and efficiency of recognition, and can adapt to new commodities and classification changes.

[0039] Other beneficial effects of the present invention will be explained through the introduction of specific technical features and technical solutions in the specific implementation methods. Those skilled in the art should be able to understand the beneficial technical effects brought about by the introduction of these technical features and technical solutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments with reference to the following drawings:

[0041] Figure 1 Flow chart of the method of the present invention.

[0042] Figure 2 This is a flowchart of an embodiment of the present invention. DETAILED DESCRIPTION

[0043] The present invention will be described in detail below with reference to specific embodiments. The following examples will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those skilled in the art, several changes and improvements can be made without departing from the scope of the present invention. These all fall within the scope of protection of the present invention.

[0044] Reference Figure 1 As shown, an intelligent image quality detection system includes:

[0045] Image Data Entry and Synchronization Module: This module supports automatic batch import of image data from different supplier systems and synchronizes updates. It uses an efficient data transmission protocol to ensure the integrity and consistency of image information and avoid data loss and duplication.

[0046] Original Image Storage Module: This module provides a distributed storage solution to ensure secure storage and fast access to original images. It leverages redundant storage technology to ensure the persistence and reliability of image data, providing basic data support for subsequent quality testing.

[0047] Automatic detection module: This module uses a deep learning model to detect image quality and automatically identifies quality issues in images, including blur, size, proportion, background color, etc. This module accelerates computing and achieves efficient parallel processing, improving detection speed and accuracy. The recognition process is shown in the following table:

[0048]

[0049] Rejected Image Identification and Cause Analysis Module: This module categorizes detected rejected images and generates a rejection analysis report. The report includes the type of problem for each image, helping users quickly locate and understand image quality defects. By using this module, the system can extract valuable information from inspection results for trend analysis and forecasting.

[0050] Image optimization module: provides intelligent image optimization tools, using intelligent image technology to improve image quality to achieve the best display effect.

[0051] See the table below for details:

[0052] technology Specific methods Main objectives Key Mechanisms Automatic white balance Grayscale world, white point hypothesis Correcting color casts Estimated lighting based on the scene's average / brightest pixel Noise reduction Median filter, Gaussian filter Noise Removal Pixel neighborhood median / weighted average Sharpen Kernel convolution Enhance Edges / Details High-pass filter kernel Smart Crop / Zoom Seam cutting Content-Aware Resizing Remove / Insert Low Energy Pixel Path Smart Fill GANs Fill in missing areas appropriately Generative models learn image statistics / context

[0053] Qualified image storage module: Qualified images that have undergone quality inspection and optimization are stored in an efficient image database, supporting fast retrieval and distribution to meet the e-commerce platform's demand for high-quality images.

[0054] Machine Learning Results Processing Module: This module continuously analyzes detection and optimization results using machine learning algorithms, optimizes model parameters, and improves the accuracy and efficiency of the detection system. This module automatically updates the learning model through a feedback mechanism to adapt to different image data and quality standards. Leveraging machine learning algorithms, the system continuously optimizes the detection model and improves detection accuracy.

[0055] Each module works together to achieve automated image quality detection and product identification. The system adopts a cloud computing architecture, utilizing distributed computing and storage technologies to support parallel processing of large-scale images, ensuring system efficiency and scalability.

[0056] By integrating these modules, the system significantly improves the efficiency and accuracy of image quality testing, helping companies optimize product display strategies and enhance the customer shopping experience, thereby strengthening the competitive advantage of e-commerce platforms. The system features an automated processing pipeline that automates all steps from image upload, pre-processing, quality testing, to result output, reducing manual intervention and improving processing efficiency.

[0057] Reference Figure 2 As shown, the present invention also provides an intelligent image quality detection method, which aims to improve the efficiency and accuracy of product identification and is particularly suitable for e-commerce platforms and inventory management systems. The method uses deep learning technology to automatically learn the relationship between images and products, thereby achieving accurate product identification and classification verification. The specific steps include:

[0058] Step 1: Synchronize images, products, and categories:

[0059] The system synchronizes product images and their corresponding product information and classification data from different suppliers to ensure the real-time and consistency of data, providing an accurate basis for subsequent processing.

[0060] Step 2: Automatically detect and classify:

[0061] Use a deep learning model to automatically detect and identify product categories. The model analyzes features in the image and matches them with classification information in the database to provide preliminary classification results.

[0062] Step 3: Recommend products and categories after identifying errors:

[0063] For products that are misidentified or uncertain, the system recommends products and categories based on the model. The model's recommendations are based on learning and analyzing historical data, providing reasonable classification options for users to consider.

[0064] Step 4: Manual verification and judgment:

[0065] Users manually verify the products and categories recommended by the system to determine their accuracy. Manual verification serves as a validation step for the system's recognition results to ensure the accuracy of the classification.

[0066] Step 5: Manually adjust products and categories:

[0067] Based on the manual verification results, users can adjust and correct products and categories to ensure the accuracy and consistency of the final data.

[0068] Step 6: Storing qualified products and classified data:

[0069] The qualified products and classification data that have been verified and adjusted are stored in the database and managed and used as high-quality product information.

[0070] Step 7: Machine Learning Qualified Data:

[0071] The system uses qualified product and classification data to continuously train and optimize machine learning models, continuously improving recognition accuracy and efficiency and adapting to new product and classification changes.

[0072] The present invention achieves efficient recognition and classification of product images by combining automatic learning and manual verification, thereby improving the product management capabilities and user experience of the e-commerce platform.

[0073] The above is a basic embodiment of the present invention. The technical solution of the present invention is further described below through a preferred embodiment.

[0074] Example 1

[0075] An intelligent image quality detection method, comprising:

[0076] 1. Image data collection and preprocessing:

[0077] Collection: Collect product images from different suppliers to ensure the diversity of data sources.

[0078] Preprocessing: De-noising, standardization, resizing, and format conversion are performed on the images to ensure the consistency and high quality of the input data, laying the foundation for subsequent processing.

[0079] 2. Efficient and intelligent image quality detection method

[0080] Automatic detection: This utilizes deep learning technology, using deep learning models for feature extraction and classification to perform image quality checks. These models are pre-trained and fine-tuned to effectively identify image quality issues such as blur and overexposure. Fine-tuning requires training at the image block level and may require the design of a specialized aggregation layer to combine block-level predictions to generate a quality assessment for the entire image. This involves unfreezing the top layer of the base model, recompiling, and continuing the training cycle. This includes head training, which is training for a specific classification or regression objective.

[0081] Quality Scoring and Labeling: This uses computer vision technology, including image segmentation, edge detection, and feature point matching, to perform detailed image analysis and processing. These techniques enable the system to accurately locate image defects and assess their impact, generating a quality score for each image and labeling specific quality issues. This information guides subsequent optimization decisions.

[0082] 3. Product and classification identification methods

[0083] Automatic Learning and Feature Extraction: Through automatic learning mechanisms, key features in product images are extracted and associated with product classification information. Utilizing transfer learning and adaptive learning techniques, the model can quickly adapt to the recognition needs of new products and new categories.

[0084] Product Image Key Feature Extraction: This process converts the raw product image pixel data into a compact and informative feature vector that captures the image's key visual attributes, such as shape, color, texture, style, and object composition. The input product image undergoes necessary preprocessing, such as resizing to the model's required size and normalizing pixel values, before being fed into a pre-trained CNN image encoder (the encoder output is the image's feature vector).

[0085] Product classification information: The product classification data label is based on the original classification data. The data set is trained according to the characteristics of products in the same category. The degree of correlation is judged by the distance (similarity) between the image features and the product classification.

[0086] Product Identification and Classification: After an image is uploaded or scanned, the system automatically identifies the product and categorizes it into the appropriate category. The identification results are compared with the product database to ensure accurate classification.

[0087] 4. Identification and feedback of unqualified images

[0088] Identification and reporting: The system identifies substandard images and generates a quality report, pointing out specific areas that need improvement.

[0089] Feedback mechanism: Feedback unqualified images and their problems to users, and suggest optimization.

[0090] 5. Image optimization and adjustment

[0091] Optimization Tools: Provides automated and manual optimization tools to help users adjust image quality. Optimization measures include denoising, color correction, sharpening, and contrast adjustment.

[0092] 6. Qualified Data Storage and Management

[0093] Storage and Classification: Store and manage optimized, qualified images and classification results. Images are stored in a distributed database, supporting fast retrieval and distribution.

[0094] 7. Machine Learning Model Optimization

[0095] Continuous Learning: Utilize qualified data for ongoing model training and optimization, continuously improving the accuracy and efficiency of detection and recognition through feedback mechanisms and online learning.

[0096] 8. User Interface and Interaction Design

[0097] User Interaction: An intuitive user interface is designed, providing straightforward operation processes and detailed inspection reports, allowing users to easily view inspection results, quality scores, identification classifications, and optimization suggestions. The interface supports multi-language and multi-platform operations to enhance the user experience.

[0098] The present invention can provide an efficient, accurate and intelligent solution for image quality detection, and offers strong technical support for e-commerce platforms and enterprises.

[0099] Those skilled in the art will appreciate that, in addition to implementing the system and its various devices, modules, and units provided by the present invention in purely computer-readable program code, it is entirely possible to implement the same functions of the system and its various devices, modules, and units provided by the present invention in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system and its various devices, modules, and units provided by the present invention can be considered a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; the devices, modules, and units for implementing various functions can also be considered as both software modules implementing the method and structures within the hardware component.

[0100] The above describes specific embodiments of the present invention. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art may make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. The embodiments of this application and the features in the embodiments may be combined with each other in any manner unless there is a conflict.

Claims

1. An intelligent image quality detection system, characterized in that: include: Image data entry and synchronization module, used to import and update original image data; Original picture storage module, used to store original picture data; An automatic detection module is used to detect the original image data and classify it into qualified images and unqualified images; Unqualified image recognition and cause analysis module, used to analyze and process the reasons for unqualified images; The image optimization module is used to optimize the image quality of unqualified images and transfer the optimized images to the qualified image storage module; The qualified image storage module is used to store optimized images and qualified images, and organize and summarize them into qualified data sets; The machine learning processing result module uses qualified data sets to continuously train and optimize the model, and uses the optimized model to detect, identify and classify the original images.

2. The intelligent image quality detection system according to claim 1, characterized in that: Also includes: User interaction module, used to display detection results, quality scores, identification classification and optimization suggestions; The user makes manual corrections and adjustments based on the recognition results of the image to complete the interaction with the system.

3. The intelligent image quality detection system according to claim 1, characterized in that: The picture data entry and synchronization module can automatically import picture data in batches from different supplier systems, update them synchronously, and use a data transmission protocol to maintain picture information.

4. The intelligent image quality detection system according to claim 1, characterized in that: The original image storage module includes the use of redundant storage technology and provides basic data support for image quality detection.

5. The intelligent image quality detection system according to claim 1, characterized in that: The automatic detection module performs image quality detection based on a deep learning model and automatically identifies image quality problems; the image quality problems include image blurring problems, image size problems, image proportion problems, and image background color problems.

6. The intelligent image quality detection system according to claim 1, characterized in that: The unqualified image recognition and cause analysis module can classify the problems of unqualified images and generate an unqualified analysis report.

7. The intelligent image quality detection system according to claim 1, characterized in that: The machine learning processing result module continuously analyzes the detection and optimization results through machine learning algorithms, optimizes the parameters of the learning model, and automatically updates the learning model through a feedback mechanism.

8. An intelligent picture quality detection method, based on the intelligent picture quality detection system according to any one of claims 1 to 7, characterized in that: include: Step S1: Collect synchronized product images and their corresponding product information and classification data from different suppliers; Step S2: pre-processing the collected product images and their corresponding product information and classification data to obtain pre-processed results; the pre-processing includes denoising, standardization, size adjustment, and format conversion of the product images; Step S3: Using a deep learning model to automatically detect and identify the category of the product after preprocessing the product image to obtain the recognition result; Step S4: Classify the product images into qualified images and unqualified images according to the recognition results; Step S5: Use the deep learning model to make improvement suggestions for unqualified images; Step S6: Manually check the unqualified images and make further adjustments to the products and categories; Step S7: Storing qualified images and adjusted and corrected images so that the system can continuously train and optimize the machine learning model based on these images; Step S8: Use the optimized machine learning model to identify and detect product images.

9. The intelligent image quality detection method according to claim 8, characterized in that: The step S3 includes the following sub-steps: Step S3.1: Use a deep learning model to extract product image features and perform quality inspection; Step S3.2: Generate a quality score for each product image and mark specific quality issues.

10. The intelligent image quality detection method according to claim 8, characterized in that: Use automatic learning mechanisms to extract key features from product images and associate them with product classification information; use transfer learning and adaptive learning techniques to adapt the model to the recognition needs of new products and new classifications.

Images