A cloud storage method and system for digital information

Abstract

A cloud storage method and system for digital information, characterized by comprising: acquiring digital image data and performing image preprocessing; enhancing image quality, improving visual perception or improving the performance of processing algorithms; identifying and extracting key features in images, dividing images into different regions or objects and analyzing them; identifying and classifying targets or patterns in images, and if the image contains moving objects, tracking the objects to understand the movement trajectory of the objects; further processing the results, analyzing the processing results, evaluating the algorithm performance, and presenting the results in a visual manner; optimizing and adjusting the algorithms as needed, and recording the entire processing flow, parameter settings, and experimental results. The present invention realizes accurate recognition and classification of targets in images through image processing technology, improves the intelligence and accuracy of the system, and efficient image processing methods can process image data more quickly, reduce delays, and enhance real-time performance.

Description

Technical Field

[0001] The present invention belongs to the technical field of digital information cloud storage, and in particular, relates to a digital information cloud storage method and system. Background Art

[0002] With the widespread application of image processing technology, issues involving personal privacy have become more prominent. How to balance performance and privacy in image processing is an important consideration. Image processing algorithms are usually designed and tested under ideal conditions, but may face various uncertainties in practical applications. Robustness is the key to dealing with these uncertainties. When the training data is relatively small, especially in some special fields, the generalization ability of the model may be limited. Small sample learning is a common challenge. For applications that require real-time processing, image processing algorithms may require higher efficiency and low latency. This is especially important for embedded systems and real-time monitoring applications. The performance of image processing algorithms may degrade under different environments and conditions. Adversarial attacks and covert attacks may have a negative impact on the performance of image processing algorithms. Therefore, developing defensive technologies against adversarial attacks and covert attacks is an important direction. Summary of the invention

[0003] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0004] In view of the problems existing in the above or existing cloud storage methods for digital information, the present invention is proposed.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] In a first aspect, an embodiment of the present invention provides a cloud storage method for digital information, including: acquiring digital image data and performing image preprocessing; enhancing image quality, improving visual perception or improving the performance of a processing algorithm; identifying and extracting key features in an image, dividing the image into different regions or objects and analyzing them; identifying and classifying targets or patterns in an image, and if the image contains a moving object, tracking the object to understand the movement trajectory of the object; further processing the results, analyzing the processing results, evaluating the algorithm performance, and presenting the results in a visual manner; optimizing and adjusting the algorithm as needed, and recording the entire processing flow, parameter settings, and experimental results.

[0007] As a preferred solution of the cloud storage method for digital information described in the present invention, the step of acquiring digital image data and performing image preprocessing includes: acquiring a digital image through a camera or a sensor, and if the image is in color, converting it into a grayscale image; using a filter to smooth the image and remove noise, using a mean filter to smooth the image, enhancing the image contrast through histogram equalization, enhancing the image edge by applying an edge enhancement filter, and converting the image into a binary image by setting a threshold.

[0008] As a preferred solution of the cloud storage method for digital information described in the present invention, the enhanced image quality, improved visual perception or improved performance of the processing algorithm includes: adjusting the colors in the image to correct color deviations caused by lighting conditions or sensor problems, and if the image is taken under foggy or atmospheric interference conditions, using defogging technology to reduce the haze effect.

[0009] As a preferred solution of the cloud storage method for digital information of the present invention, the step of identifying and extracting key features in an image, dividing the image into different regions or objects and analyzing the images includes:

[0010] Use Harris corner detection to extract the color, texture, shape and edge of the image; divide the image into different regions or objects, divide the image into two or more regions based on the threshold of the pixel gray value, perform shape analysis on each segmented region, obtain information about its geometric structure and features, match the extracted features with predefined models or templates to identify and classify objects in the image, compare the regions in the image with known objects or categories, and use machine learning algorithms to identify the actual objects in the image. The expression is as follows:

[0011] Among them, w is the normal vector of the hyperplane, αi is the Lagrange multiplier, yi is the category label, and xi is the feature vector;

[0012]

[0013] Among them, b is the intercept of the hyperplane, yi is the sample, xi is the feature vector of the sample, xk is the support vector on the boundary, and yk is the true label;

[0014] The segmentation results are further processed and marked.

[0015] As a preferred solution of the cloud storage method for digital information of the present invention, the identification and classification of targets or patterns in the image includes: extracting color, texture and shape features for target identification and classification from the image, training a classifier using a convolutional neural network so that it can distinguish different target categories, and applying the trained classifier in the image to detect and identify the position of the target object. The convolutional neural network expression is as follows:

[0016] Train the classifier:

[0017]

[0018] Where Xi is the i-th image in the training set, Yi is the corresponding true label, and F(⋅;W) represents the output of the model;

[0019] Object Detection:

[0020] After the training is completed, the trained classifier is applied to the new image Xnew to detect and identify the location of the target object;

[0021]

[0022] Where Y is the predicted target category and F(Xnew;W) is the output of the model for new images.

[0023] As a preferred solution of the cloud storage method for digital information described in the present invention, if the image contains a moving object, object tracking is performed to understand the movement trajectory of the object, including: in the first frame, the position and bounding box of the target are initialized using the result of target recognition, for each frame, the features of the target are extracted, and then the target is matched using feature matching technology; the position of the target in the current frame is obtained by estimating the movement of the target features, and the bounding box of the target is updated according to the result of the motion estimation; the movement trajectory of an object is maintained and the position of the target in each frame is recorded; the movement trajectory of the target is analyzed, and the movement pattern, speed and acceleration of the target are understood using computer vision or machine learning methods, and in the process of trajectory analysis, anomaly detection is performed to identify abnormal behavior of the target or irregular patterns in the trajectory.

[0024] As a preferred solution of the cloud storage method for digital information of the present invention, the further processing of the results, analyzing the processing results, evaluating the algorithm performance, and presenting the results in a visual manner include: in target detection, there are multiple overlapping bounding boxes, using non-maximum suppression to select the box with the highest confidence, and suppressing the overlapping boxes with lower confidence; calculating the accuracy and recall rate of target detection, the accuracy rate represents the ability of the model to correctly identify the target, and the recall rate represents the ability of the model to find all targets, and the expression is as follows:

[0025]

[0026] Construct a confusion matrix to display the number of true positives and false positives and true negatives and false negatives, and analyze the performance of the model; the F1 score is the harmonic mean of precision and recall, and the performance of the algorithm is evaluated by the F1 score, which is expressed as follows:

[0027]

[0028] Visualize the results of object detection by drawing bounding boxes on the image to show the performance of the algorithm. Adjust the learning rate and threshold hyperparameters of the model based on the results of the performance evaluation.

[0029] In a second aspect, an embodiment of the present invention provides a cloud storage system for digital information, which includes an image data acquisition and processing module for acquiring digital image data and performing image preprocessing; an image enhancement module for enhancing image quality, improving visual perception or improving the performance of a processing algorithm; an image feature extraction and analysis module for identifying and extracting key features in an image, dividing the image into different regions or objects and performing analysis; an image recognition and classification module for identifying and classifying targets or patterns in an image, and if the image contains a moving object, performing object tracking to understand the object's motion trajectory; an evaluation and optimization module for further processing the results, analyzing the processing results, evaluating the algorithm performance, and presenting the results in a visual manner; optimizing and adjusting the algorithm as needed, and recording the entire processing flow, parameter settings, and experimental results.

[0030] In a third aspect, an embodiment of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the processor executes the computer program, any step of the above-mentioned cloud storage method for digital information is implemented.

[0031] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the above-mentioned cloud storage method for digital information is implemented.

[0032] The beneficial effects of the present invention are as follows: the present invention realizes accurate recognition and classification of targets in images through image processing technology, thereby improving the intelligence and accuracy of the system. For real-time applications, efficient image processing methods can process image data more quickly, reduce latency, and enhance real-time performance. This image processing method can handle complex scenes, overcome problems such as lighting changes, occlusion, and noise, and will improve the adaptability of the system in the real world. By considering changes and uncertainties in various environments, the image processing method is made more robust, improving the stability and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0034] Figure 1 A flowchart of a cloud storage method for digital information provided by an embodiment of the present invention.

[0035] Figure 2 A schematic diagram of the structure of a cloud storage system for digital information provided by an embodiment of the present invention.

[0036] Figure 3 A flow chart of the steps of image motion trajectory analysis of a cloud storage method for digital information provided by an embodiment of the present invention.

[0037] Figure 4 An internal structural diagram of a computer device of a cloud storage method and system for digital information provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0038] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0039] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0040] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0041] Example

[0042] Reference Figure 1~Figure 3 , is an embodiment of the present invention, which provides a cloud storage method for digital information, including:

[0043] S1: Acquire digital image data and perform image preprocessing.

[0044] Preferably, a digital image is acquired through a camera or sensor, and if the image is in color, it is converted into a grayscale image; a filter is used to smooth the image and remove noise, a mean filter is used to smooth the image, the image contrast is enhanced by histogram equalization, the image edge is enhanced by applying an edge enhancement filter, and the image is converted into a binary image by setting a threshold.

[0045] Furthermore, the process of converting the image into a binary image by thresholding is as follows:

[0046] import cv2

[0047] import numpy as np

[0048] from matplotlib import pyplot as plt

[0049] #Read color image

[0050] image = cv2.imread('example_image.jpg', cv2.IMREAD_GRAYSCALE)

[0051] #Set the threshold. Here we use Otsu's binarization method to automatically calculate the threshold_, binary_image = cv2.threshold(image, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)

[0052] # Display the original image and binary image

[0053] plt.subplot(1, 2, 1), plt.imshow(image, cmap='gray')

[0054] plt.title('Original Image'), plt.xticks([]), plt.yticks([])

[0055] plt.subplot(1, 2, 2), plt.imshow(binary_image, cmap='gray')

[0056] plt.title('Binary Image'), plt.xticks([]), plt.yticks([])

[0057] plt.show().

[0058] S2: Enhance the quality of images, improve visual perception or increase the performance of processing algorithms.

[0059] Preferably, the colors in the image are adjusted to correct color deviations caused by lighting conditions or sensor problems. If the image is taken under foggy or atmospheric interference conditions, defogging technology is used to reduce the haze effect.

[0060] S3: Identify and extract key features in the image, divide the image into different regions or objects and analyze them.

[0061] Preferably, Harris corner detection is used to extract the color, texture, shape and edge of the image; the image is divided into different regions or objects, based on the threshold of the pixel gray value, the image is divided into two or more regions, each segmented region is subjected to shape analysis, information about its geometric structure and features is obtained, the extracted features are matched with a predefined model or template to identify and classify the objects in the image, the regions in the image are compared with known objects or categories, and the actual objects in the image are identified using a machine learning algorithm, the expression of which is as follows:

[0062] Among them, w is the normal vector of the hyperplane, αi is the Lagrange multiplier, yi is the category label, and xi is the feature vector;

[0063]

[0064] Among them, b is the intercept of the hyperplane, yi is the sample, xi is the feature vector of the sample, xk is the support vector on the boundary, and yk is the true label;

[0065] The segmentation results are further processed and marked.

[0066] Furthermore, the Harris corner detection method is used to detect corners in the image. These corners represent significant local features in the image, such as ridges, corners, or edges. The color image is converted to the HSV color space to separate the color information and highlight the color differences between different areas in the image, such as the blue of the lake and the green of the trees. Texture extraction methods such as Canny edge detection are used to highlight the texture information in the image, which helps to identify the details of the trees or the outlines of the mountains. Based on the threshold of the pixel gray value, the image is divided into two regions, one containing pixels with high gray values ​​and the other containing pixels with low gray values, dividing the image into distinct black and white parts. Shape analysis is performed on each segmented region, and the shapes of lakes and mountains are identified by finding the outlines, and these regions are marked in the image.

[0067] S4: Identify and classify objects or patterns in images.

[0068] Preferably, color, texture and shape features for target recognition and classification are extracted from the image, a classifier is trained using a convolutional neural network so that it can distinguish different target categories, and the trained classifier is applied to the image to detect and identify the location of the target object. The convolutional neural network expression is as follows:

[0069] Train the classifier:

[0070]

[0071] Where Xi is the i-th image in the training set, Yi is the corresponding true label, and F(⋅;W) represents the output of the model;

[0072] Object Detection:

[0073] After the training is completed, the trained classifier is applied to the new image Xnew to detect and identify the location of the target object;

[0074]

[0075] Where Y is the predicted target category and F(Xnew;W) is the output of the model for new images.

[0076] Furthermore, collect a dataset of images containing different target categories and ensure that each image has a corresponding label. The dataset should contain images with large variations in color, texture, and shape. Use image processing techniques to extract the color, texture, and shape features of the image. For example, observe the distribution of colors, use texture descriptors (such as Canny edge detection), and detect the main shapes in the image. Design a convolutional neural network (CNN) model that accepts images as input and outputs probability distributions of different target categories. Convolutional layers help capture spatial features in images, while fully connected layers can handle combinations of features. Select a loss function, optimizer, and evaluation metric and compile the model. Then, train the model using the training dataset so that it can learn to extract image features and classify targets. Use the trained model for target detection. Input the new image into the model to obtain the probability distribution of each target category. Determine the presence or absence of the target by setting a threshold.

[0077] S5: If the image contains a moving object, the object is tracked to understand the moving trajectory of the object.

[0078] Preferably, in the first frame, the position and bounding box of the target are initialized using the result of target recognition, and for each frame, the features of the target are extracted, and then the target is matched using feature matching technology; the position of the target in the current frame is obtained by estimating the motion of the target features, and the bounding box of the target is updated according to the result of the motion estimation; the motion trajectory of an object is maintained and the position of the target in each frame is recorded; the motion trajectory of the target is analyzed, and computer vision or machine learning methods are used to understand the motion pattern, speed and acceleration of the target, and in the process of trajectory analysis, anomaly detection is performed to identify abnormal behavior of the target or irregular patterns in the trajectory.

[0079] Furthermore, suppose that a vehicle in an image is to be identified and processed. In the first frame, the vehicle in the image is detected by a target recognition method. The center of the vehicle is selected as the initial position, and the color histogram and texture features of the vehicle are obtained. In subsequent frames, the color histogram and texture features of the vehicle area in the current frame are extracted. The current features are matched with the initial features using feature matching techniques, such as histogram matching or feature point matching. The direction and speed of movement of the vehicle are estimated using the optical flow method. Based on the results of the motion estimation, the bounding box of the vehicle is updated. The position of the vehicle in each frame is recorded and the movement trajectory of the vehicle is maintained. The speed and acceleration of the vehicle are calculated using the trajectory information. Machine learning algorithms or statistical methods are applied for anomaly detection to identify possible abnormal behaviors or irregular motion patterns.

[0080] S6: Further process the results, analyze the processing results, evaluate the algorithm performance, and present the results in a visual way.

[0081] Preferably, in target detection, there are multiple overlapping bounding boxes, and non-maximum suppression is used to select the box with the highest confidence, and suppress the overlapping boxes with lower confidence; the accuracy and recall of target detection are calculated, the accuracy indicates the ability of the model to correctly identify the target, and the recall indicates the ability of the model to find all targets, and its expression is as follows:

[0082]

[0083] Construct a confusion matrix to display the number of true positives and false positives and true negatives and false negatives, and analyze the performance of the model; the F1 score is the harmonic mean of precision and recall, and the performance of the algorithm is evaluated by the F1 score, which is expressed as follows:

[0084]

[0085] Visualize the results of object detection by drawing bounding boxes on the image to show the performance of the algorithm. Adjust the learning rate and threshold hyperparameters of the model based on the results of the performance evaluation.

[0086] Furthermore, consider an object detection task that contains 10 traffic signs. Each sign has a true bounding box and a bounding box predicted by the model, as well as a confidence score for each bounding box.

[0087] True bounding box 1: (50, 100, 120, 180) Predicted bounding box 1: (55, 95, 125, 185), confidence score: 0.8 True bounding box 2: (200, 50, 280, 120) Predicted bounding box 2: (210, 45, 290,125), confidence score: 0.7

[0088] Selected bounding boxes after NMS: Selected bounding box 1: (55, 95, 125, 185), confidence score: 0.8 Selected bounding box 2: (210, 45, 290, 125), confidence score: 0.7.

[0089] Compute the confusion matrix:

[0090] True examples (1 Positive): 8

[0091] False positives (0 Positive): 2

[0092] True negative examples (1 Negative): 0

[0093] False negatives (0 Negative): 0

[0094] Calculate the accuracy, recall, and F1 score: Accuracy: 0.8 Recall: 1.0 F1 score: 0.8889, the learning rate is adjusted to 0.0001, the threshold is adjusted to 0.6, and the performance of the algorithm is demonstrated by drawing the selected bounding box on the image.

[0095] In a preferred embodiment, a cloud storage system for digital information includes an image data acquisition and processing module for acquiring digital image data and performing image preprocessing; an image enhancement module for enhancing image quality, improving visual perception or improving the performance of a processing algorithm; an image feature extraction and analysis module for identifying and extracting key features in an image, dividing the image into different regions or objects and performing analysis; an image recognition and classification module for identifying and classifying targets or patterns in an image, and if the image contains a moving object, performing object tracking to understand the object's motion trajectory; an evaluation and optimization module for further processing the results, analyzing the processing results, evaluating the algorithm performance, and presenting the results in a visual manner; optimizing and adjusting the algorithm as needed, and recording the entire processing flow, parameter settings, and experimental results.

[0096] The above-mentioned unit modules may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above-mentioned modules.

[0097] In one embodiment, a computer device is provided, which may be a terminal, and includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0098] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A cloud storage method for digital information, characterized in that: include: Acquire digital image data and perform image preprocessing; Enhance the quality of images, improve visual perception or increase the performance of processing algorithms; Identify and extract key features in images, divide images into different areas or objects and analyze them; Identify and classify objects or patterns in the image. If the image contains moving objects, track the objects to understand their motion trajectory. Further process the results, analyze the processing results, evaluate the algorithm performance, and present the results in a visual way; Optimize and adjust the algorithm as needed, and record the entire processing flow, parameter settings, and experimental results; The identifying and extracting key features in the image, dividing the image into different regions or objects and analyzing them, includes: Use Harris corner detection to extract the color, texture, shape and edge of the image; divide the image into different regions or objects, divide the image into two or more regions based on the threshold of the pixel gray value, perform shape analysis on each segmented region, obtain information about its geometric structure and features, match the extracted features with predefined models or templates to identify and classify objects in the image, compare the regions in the image with known objects or categories, and use machine learning algorithms to identify the actual objects in the image. The expression is as follows: Where ω is the normal vector of the hyperplane, α i is the Lagrange multiplier, y i is the class label, x i is the feature vector of the sample; Where b is the intercept of the hyperplane, y i is the category label, is the transpose of the sample’s eigenvector, x k is the support vector on the boundary, y k is the true label; Further process the segmentation results and mark the segmentation results; Harris corner detection method is used to detect corners in the image, which represent significant local features in the image. Based on the threshold of pixel grayscale value, the image is divided into two regions, one containing pixels with high grayscale value and the other containing pixels with low grayscale value, which divides the image into obvious black and white parts. Shape analysis is performed on each segmented region. The identifying and classifying objects or patterns in an image comprises: Color, texture and shape features for target recognition and classification are extracted from the image, a classifier is trained using a convolutional neural network to enable it to distinguish different target categories, and the trained classifier is applied to the image to detect and identify the category of the target object. The convolutional neural network expression is as follows: Train the classifier: Among them, X i is the i-th image in the training set, Y i is the corresponding true label, F(X i ; W) represents the output of the model; Object Detection: After the training is completed, the trained classifier is applied to the new image Xnew to detect and identify the category of the target object; Y=argmax Y F(X new ;W) Among them, Y is the predicted target category, F(X new ; W) is the output of the model for new images; Collect an image dataset containing different target categories, ensure that each image has a corresponding label, the dataset contains images with changes in color, texture, and shape, use image processing technology to extract the color, texture, and shape features of the image, observe the distribution of color, use texture descriptors and detect the main shapes in the image, design a convolutional neural network model, which accepts images as input and outputs probability distributions of different target categories. The convolution layer helps to capture spatial features in the image, while the fully connected layer can process the combination of features, select the loss function, optimizer, and evaluation indicators, compile the model, train the model using the training dataset, so that it learns to extract image features and classify targets, use the trained model for target detection, input the new image into the model, obtain the probability distribution of each target category, and determine the presence or absence of the target by setting a threshold.

2. The cloud storage method for digital information according to claim 1, characterized in that: The step of acquiring digital image data and performing image preprocessing includes: Acquire a digital image through a camera or sensor. If the image is in color, convert it to a grayscale image. Use filters to smooth the image and remove noise. Use mean filtering to smooth the image. Enhance the image contrast through histogram equalization. Enhance the image edges by applying edge enhancement filters. Convert the image to a binary image by setting a threshold.

3. The cloud storage method for digital information according to claim 1, characterized in that: If the image contains a moving object, object tracking is performed to understand the moving trajectory of the object, including: In the first frame, the target position and bounding box are initialized using the target recognition results. For each frame, the target features are extracted, and then feature matching technology is used to match the target. The position of the target in the current frame is obtained by estimating the motion of the target features, and the target bounding box is updated according to the results of the motion estimation. The motion trajectory of an object is maintained and the position of the target in each frame is recorded. The motion trajectory of the target is analyzed, and computer vision or machine learning methods are used to understand the target's motion pattern, speed, and acceleration. During the trajectory analysis process, anomaly detection is performed to identify abnormal behaviors of the target or irregular patterns in the trajectory.

4. The cloud storage method for digital information according to claim 1, characterized in that: The further processing of the results, analysis of the processing results, evaluation of the algorithm performance, and presentation of the results in a visual manner include: In target detection, there are multiple overlapping bounding boxes. Non-maximum suppression is used to select the box with the highest confidence and suppress the overlapping boxes with lower confidence. The accuracy and recall of target detection are calculated. The accuracy indicates the ability of the model to correctly identify the target, and the recall indicates the ability of the model to find all targets. The expressions are as follows: Construct a confusion matrix to display the number of true positives, false positives, and false negatives, and analyze the performance of the model; the F1 score is the harmonic mean of precision and recall. The performance of the algorithm is evaluated by the F1 score, and its expression is as follows: Among them, Precision is the accuracy of the model; Recall is the recall rate of the model; Visualize the results of object detection by drawing bounding boxes on the image to show the performance of the algorithm. Adjust the learning rate and threshold hyperparameters of the model based on the results of the performance evaluation.

5. A cloud storage system for digital information, comprising the cloud storage method for digital information according to any one of claims 1 to 4, characterized in that: include: An image data acquisition and processing module is used to acquire digital image data and perform image preprocessing; Image enhancement module, used to enhance the quality of images, improve visual perception or improve the performance of processing algorithms; Image feature extraction and analysis module, used to identify and extract key features in images, divide images into different areas or objects and analyze them; Image recognition and classification module, used to identify and classify targets or patterns in images. If the image contains moving objects, it will track the objects and understand their motion trajectory. Evaluate and optimize the module, further process the results, analyze the processing results, evaluate the algorithm performance, and present the results in a visual way; optimize and adjust the algorithm as needed, and record the entire processing flow, parameter settings, and experimental results.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the cloud storage method for digital information described in any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the cloud storage method for digital information described in any one of claims 1 to 4 are implemented.

Images