Intelligent system and method for whole-course tracking and dynamic evaluation of study literacy of pupils

Abstract

The invention discloses an intelligent system for whole-course tracking and dynamic evaluation of study literacy of pupils. The system comprises an acquisition module for acquiring image data and text data; the preprocessing module preprocesses the image data; the feature matching module performs feature extraction on the preprocessed image by adopting a SIFI algorithm, performs feature matching by adopting an improved FLANN algorithm, and identifies experimental operation behaviors of students; the text processing module performs natural language processing on the text data; the model processing module performs quantitative analysis on the experimental operation behavior and the processed information by using a quality evaluation model to obtain evaluation data; the analysis module adopts a time sequence analysis method to analyze the time sequence data set to obtain the learning performance trend of the student; and the graph display module displays the evaluation data and the learning performance trend in a graph form. Various types of data are automatically collected and processed, comprehensive, accurate and objective evaluation of student research literacy is realized, and the performance of students can be dynamically tracked in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent assessment of student literacy, and in particular to an intelligent system and method for full-process tracking and dynamic evaluation of primary school students' research literacy. Background Art

[0002] At present, with the development of educational informatization, the cultivation and evaluation of students' research literacy has received increasing attention. However, the existing evaluation methods mostly rely on traditional subjective evaluation methods such as written tests and teacher observations, which have the following main problems:

[0003] Inconsistent evaluation standards: There are differences in the evaluation standards for research literacy in different regions and schools, which makes it difficult to compare the evaluation results and objectively reflect the students' true level.

[0004] The evaluation method is single: it mainly relies on the subjective judgment of teachers, lacks scientific quantitative evaluation methods, and is difficult to fully reflect students' research literacy.

[0005] Insufficient data collection and analysis capabilities: Collecting student performance data through classroom observation, homework grading, written tests, etc. Traditional methods make it difficult to collect and analyze student learning data in real time and comprehensively, and fail to provide timely and effective feedback for teaching improvements. Summary of the invention

[0006] In view of the defects in the prior art, the embodiments of the present invention provide an intelligent system and method for full-process tracking and dynamic evaluation of primary school students' research literacy to solve the problems existing in the prior art.

[0007] In the first aspect, an embodiment of the present invention provides an intelligent system for tracking and dynamically evaluating the research literacy of primary school students, including: a collection module, a preprocessing module, a feature matching module, a text processing module, a model processing module, an analysis module and a graphic display module;

[0008] The acquisition module is used to collect image data, experimental operation image data, experimental report text data and discussion and communication text data of students in their research activities;

[0009] The preprocessing module is used to preprocess the collected image data to obtain a preprocessed image;

[0010] The feature matching module is used to extract features from the preprocessed image using the SIFI algorithm, and to perform feature matching using the improved FLANN algorithm to identify the students' experimental operation behaviors;

[0011] The text processing module is used to perform natural language processing on the collected text data to obtain processed information;

[0012] The model processing module is used to use the literacy evaluation model constructed by the SVM classifier to perform quantitative analysis on the experimental operation behavior and the processed information, and obtain evaluation data of the students' experimental operation ability and report writing ability;

[0013] The analysis module is used to sort the evaluation data according to the timestamp, construct a time series data set, and analyze the time series data set using a time series analysis method to obtain the learning performance trend of the students;

[0014] The graphic display module displays the evaluation data and learning performance trends in the form of charts.

[0015] In a second aspect, an embodiment of the present invention provides a method for full-process tracking and dynamic evaluation of primary school students' research literacy, including:

[0016] Collect image data of students' research activities, experimental operation image data, experimental report text data and discussion and communication text data;

[0017] Preprocessing the collected image data to obtain a preprocessed image;

[0018] The SIFI algorithm is used to extract features from the preprocessed images, and the improved FLANN algorithm is used for feature matching to identify students' experimental operation behaviors.

[0019] Perform natural language processing on the collected text data to obtain processed information;

[0020] The literacy evaluation model constructed by SVM classifier is used to quantitatively analyze the experimental operation behavior and processed information to obtain the evaluation data of students' experimental operation ability and report writing ability;

[0021] Sort the evaluation data according to timestamps, construct a time series data set, and use time series analysis methods to analyze the time series data set to obtain the students' learning performance trends;

[0022] Present assessment data and learning performance trends in charts and graphs.

[0023] Beneficial effects of the present invention:

[0024] The embodiment of the present invention provides an intelligent system and method for tracking and dynamically evaluating the research literacy of primary school students. By collecting various data of students in research activities, preprocessing the collected image data, performing natural language processing on text data, and using the constructed SVM classifier to construct a literacy evaluation model, the experimental operation behavior and information obtained after processing are quantitatively analyzed to obtain evaluation data of students' experimental operation ability and report writing ability, that is, students' research literacy evaluation data, and then sorting the evaluation data according to timestamps to construct a time series data set. The time series analysis method is used to analyze the trend of students' learning performance, discover students' progress and deficiencies, and finally display the evaluation data and learning performance trends in the form of charts. It can automatically collect and process various types of data, realize comprehensive, accurate and objective evaluation of students' research literacy, and can dynamically track students' performance in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0026] Figure 1 A flowchart of a method for full-process tracking and dynamic evaluation of primary school students' research literacy provided by the first embodiment of the present invention is shown;

[0027] Figure 2 A structural block diagram of an intelligent system for full-process tracking and dynamic evaluation of primary school students' research literacy provided by another embodiment of the present invention is shown. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0030] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0031] It should be further understood that the term "and / or" used in the present specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0032] As used in this specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0033] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application should have the common meanings understood by those skilled in the art to which the present invention belongs.

[0034] like Figure 1 As shown, a structural block diagram of an intelligent system for full-process tracking and dynamic evaluation of primary school students' research literacy provided by another embodiment of the present invention is shown, and the system includes: an acquisition module, a preprocessing module, a feature matching module, a text processing module, a model processing module, an analysis module and a graphic display module; the acquisition module is used to collect image data, experimental operation image data, experimental report text data and discussion and communication text data of students in research activities; the preprocessing module is used to preprocess the collected image data to obtain a preprocessed image; the feature matching module is used to extract features from the preprocessed image using the SIFI algorithm, and use the improved FLANN algorithm for feature matching to identify the students' experimental operation behavior; the text processing module is used to perform natural language processing on the collected text data to obtain processed information; the model processing module is used to use the evaluation model constructed by the SVM classifier to perform quantitative analysis on the experimental operation behavior and the processed information to obtain evaluation data of the students' experimental operation ability and report writing ability; the analysis module is used to sort the evaluation data according to the timestamp, construct a time series data set, and use the time series analysis method to analyze the time series data set to obtain the students' learning performance trend; the graphic display module displays the evaluation data and the learning performance trend in the form of a chart.

[0035] The feature matching module includes a feature extraction unit, which is used to use Gaussian difference kernels of different scales to convolve with the image to obtain a Gaussian difference scale space; detect extreme points in the Gaussian difference scale space; use a sub-pixel interpolation method to continuously interpolate discrete spatial points to obtain the extreme points and the sub-pixel positions of the extreme points in the continuous space, and determine the position and scale of the key points; remove irrelevant feature points from all detected feature points; determine the direction of the feature points, and generate a SIFT feature descriptor.

[0036] The feature matching module also includes a matching unit, which uses the RANSAC algorithm to find the mapping relationship between the paired points based on the matching point pairs obtained in the previous frame image to obtain a mapping matrix; searches within the mapping point field of the previous frame matching feature points to determine whether a paired point with a minimum Euclidean distance less than a set value is found, and if so, matches are performed; if not, determines whether the trace symbols of the feature points are the same; if they are the same, uses the original FLANN algorithm for matching; if they are not the same, no matching is performed.

[0037] The text processing module includes a text cleaning unit, a word segmentation unit and a tagging unit. The text cleaning unit is used to clean the text data, remove special characters, remove stop words, and convert the letters in the text into uppercase and lowercase letters; the word segmentation unit performs word segmentation on the text and divides the text into multiple independent words; the tagging unit tags the part of speech of the words.

[0038] The model processing module includes a model building unit. The specific method of the model building unit using the SVM classifier to build a literacy evaluation model includes: determining student ability indicators based on students' experimental operation standards; collecting students' data on ability indicators for sorting and processing, and dividing the processed data into a training set and a test set; using the training set to train the SVM classifier, and using the test set to input the SVM classifier for testing to obtain an evaluation model.

[0039] The first embodiment of the present invention provides an intelligent system for tracking and dynamically evaluating the research literacy of primary school students. By collecting various data of students in research activities, preprocessing the collected image data, performing natural language processing on text data, and using the constructed SVM classifier to construct a literacy evaluation model, the experimental operation behavior and information obtained after processing are quantitatively analyzed to obtain evaluation data of students' experimental operation ability and report writing ability, that is, students' research literacy evaluation data, and then sorting the evaluation data according to timestamps to construct a time series data set. The time series analysis method is used to analyze the trend of students' learning performance, discover students' progress and deficiencies, and finally display the evaluation data and learning performance trends in the form of charts. It can automatically collect and process various types of data, realize a comprehensive, accurate and objective evaluation of students' research literacy, and can dynamically track students' performance in real time.

[0040] like Figure 2 FIG. 1 is a flowchart of a method for full-process tracking and dynamic evaluation of primary school students' research literacy provided by a first embodiment of the present invention, the method comprising the following steps:

[0041] S1: Collect image data of students in research activities, experimental operation image data, experimental report text data and discussion and communication text data;

[0042] S2: preprocessing the collected image data to obtain a preprocessed image;

[0043] S3: The SIFI algorithm is used to extract features from the preprocessed images, and the improved FLANN algorithm is used for feature matching to identify students' experimental operation behaviors;

[0044] S4: Perform natural language processing on the collected text data to obtain processed information;

[0045] S5: The literacy evaluation model constructed by the SVM classifier is used to quantitatively analyze the experimental operation behavior and the processed information to obtain the evaluation data of the students' experimental operation ability and report writing ability;

[0046] S6: Sort the evaluation data according to the timestamp, construct a time series data set, and analyze the time series data set using a time series analysis method to obtain the student's learning performance trend;

[0047] S7: Display the evaluation data and learning performance trends in the form of charts.

[0048] The above methods also include: making targeted teaching improvement suggestions based on the evaluation results, such as strengthening experimental operation training, improving report writing skills, etc.

[0049] In this embodiment, the research activities include experimental operations, research and study activities, etc. By collecting images and video data of students' behaviors in the research activities, video data during the experimental operations, text data during the communication and discussion, and text data of the submitted experimental reports, these data are preprocessed, and the SIFI algorithm is used to extract features from the images, and the improved FLANN algorithm is combined for feature matching. The evaluation model is used to identify the students' experimental operation behaviors, and the processed text data is quantitatively analyzed using the evaluation model to obtain evaluation data of students' experimental operation ability and report writing ability. The evaluation data is also sorted by time, and the time series analysis method is used to analyze the time series data to obtain the students' learning performance trends, and the obtained evaluation data and learning performance trends are displayed in the form of charts.

[0050] The above steps can achieve a comprehensive and objective evaluation of students' research literacy by automatically collecting image information, using image recognition technology to process the collected image and video data, and using natural language processing methods to perform intelligent processing on the acquired text data. Real-time tracking and evaluation of students' learning performance, trend analysis of learning performance, timely identification of students' progress and deficiencies, and graphical display of evaluation results can help educators intuitively understand students' situation.

[0051] In this embodiment, the specific method for preprocessing the collected image data includes: image denoising, image enhancement, image geometric transformation and image normalization. Image denoising is to reduce the impact of noise on the image. In this embodiment, median filtering is used: the median value in the pixel neighborhood is used as the value of the pixel. Median filtering has a good removal effect on salt and pepper noise, and can better retain the edge information of the image. Image enhancement can improve the contrast, brightness and clarity of the image, making the features in the image more obvious. Grayscale transformation: enhance the image by adjusting the grayscale value distribution of the image. For example, linear grayscale transformation can stretch or compress the grayscale range of the image to improve the contrast; logarithmic transformation and exponential transformation can enhance the details of the dark or bright parts. Histogram equalization: by adjusting the histogram of the image, the grayscale distribution is made more uniform, thereby enhancing the contrast of the image. It can automatically determine the grayscale transformation function without manual intervention. Sharpening: enhance the edge and detail information of the image to make the image clearer. Image geometric transformation is used to correct the geometric deformation of the image or perform operations such as scaling and rotating the image. Scaling: You can change the size of an image, including zooming in and out. When zooming in, interpolation is required to fill in new pixel values. Common interpolation methods include nearest neighbor interpolation, bilinear interpolation, and bicubic interpolation. Rotation: Rotate an image at a certain angle. Image normalization allows different images to have the same size, grayscale range, or other features for subsequent processing and analysis.

[0052] In this embodiment, the specific method of extracting features from the preprocessed image using the SIFI algorithm includes:

[0053] The Gaussian difference scale space is obtained by convolving the image with Gaussian difference kernels of different scales;

[0054] Detect the extreme points in the Gaussian difference scale space;

[0055] Using the sub-pixel interpolation method, the extreme points and the sub-pixel positions of the extreme points in the continuous space are obtained by continuously interpolating the discrete spatial points, and the positions and scales of the key points are determined;

[0056] Remove irrelevant feature points from all detected feature points;

[0057] Determine the direction of the feature points and generate SIFT feature descriptors.

[0058] SIFT (Scale-invariant feature transform) is an algorithm for detecting local features. This algorithm obtains features and matches image feature points by finding the descriptors of feature points in an image and their related scale and orientation. SIFT features are not only scale-invariant, but also can achieve good detection results even if the rotation angle, image brightness and other conditions are changed. The main steps include: constructing scale space, LOG approximation DOG to find key points and detect extreme points in DOG scale space, accurately locating feature points, determining the direction of feature points, and finally generating SIFT feature descriptors.

[0059] In this embodiment, the specific method of performing feature matching using the improved FLANN algorithm includes:

[0060] According to the matching point pairs obtained in the previous frame image, the RANSAC (random sample consensus) algorithm is used to find the mapping relationship between the matching points and obtain the mapping matrix;

[0061] Search within the mapping point area of ​​the previous frame matching feature point to determine whether a matching point with a minimum Euclidean distance less than the set value is found. If so, match it;

[0062] If not, determine whether the trace signs of the feature points are the same;

[0063] If they are the same, the original FLANN algorithm is used for matching;

[0064] If they are not the same, no match is performed.

[0065] In this embodiment, based on the FLANN algorithm, the random sample consensus (RANSAC) algorithm is combined to eliminate a large number of mismatched and useless feature points, thereby greatly improving the matching speed. The original RANSAC algorithm finds a model of a data set through an iterative cycle. Each iteration of RANSAC randomly selects 4 groups of feature points to calculate their mapping relationship and counts the feature points that meet the mapping. After a certain number of iterations, the model with the largest number of feature points is selected as the final model.

[0066] The specific method of performing feature matching using the improved FLANN algorithm includes the following steps:

[0067] (1) The FLANN matching algorithm requires pairwise calculation to find the best matching point pair, in which mismatched and irrelevant feature points are also involved in the calculation, while the RANSAC algorithm requires the best matching point for the feature points, so it does not need iteration to find the model with the highest probability. Neither meets the requirements of fast video image matching. It is only necessary to select 4 groups of paired feature points from the previous frame to find the mapping matrix H. The mapping relationship is shown in formula (1):

[0068]

[0069] Among them, (x', y') is the mapping point coordinates of the feature point (x, y), and H is the mapping matrix. The solution of the mapping matrix H can expand equation (1) to obtain equation (2):

[0070]

[0071] According to formula (2), we can get formula (3):

[0072]

[0073] Since the transformation matrix H can be scaled to any scale, let h 33 =1, each set of matched feature points can get two sets of equations, so four sets of equations (3) can be obtained through four sets of feature points, and the eight parameters of the matrix H can be calculated to get the matrix H. After obtaining the mapping matrix, go to step (3).

[0074] If the current frame is the first frame of the video or no matching point is found in the previous frame, skip this step and go to step (3), and directly use the original FLANN algorithm to find the matching point.

[0075] (2) In the FLANN algorithm, the prediction area of ​​the pairing point is increased according to the RANSAC algorithm, and the pairing point is searched in the local area. By calculating the predicted pairing point of the feature point p0 (x0, y0), the 30-degree area of ​​the predicted point is used as the prediction area of ​​the pairing point. First, the mapping point coordinates of p0 are calculated by formula (1), and then the pairing point is searched in the 30-degree area of ​​the mapping point first to find the pairing point with the minimum Euclidean distance in the area of ​​p0 point. Assume that the two m-dimensional feature vectors are p(x0, y0) and p0 respectively. p1 ,x p2 ,…,x pm ),q(x q1 ,x q2 ,…,x qm ), then their Euclidean distance D pq As shown in formula (4):

[0076]

[0077] There is a high probability that the best pairing point with the minimum Euclidean distance less than 100 can be found in the mapping point area of ​​p0, because the movement range of the previous and next frame images is usually relatively small. Otherwise, go to step (3) and step (4) to continue calling the original FLANN algorithm to search for the remaining feature points.

[0078] (3) Before performing FLANN algorithm matching, the trace symbols of the feature points are classified to further screen the matching points and reduce the amount of matching calculations. Determine whether the trace symbols of the two feature points are the same. If the symbols are the same, proceed to the next step. Otherwise, skip the Euclidean distance calculation and directly judge them as different points, as shown in formula (5):

[0079]

[0080] (4) The core of the FLANN algorithm is to calculate the Euclidean distance between two feature points. By judging whether the ratio of the minimum Euclidean distance to a SURF feature point and the second minimum Euclidean distance is lower than a certain threshold, we can filter whether there is a unique matching point for the feature point. We can further filter the feature point matching pairs by formula (6):

[0081]

[0082] Where: D pq is the minimum Euclidean distance from the feature point p, D pk is the second smallest Euclidean distance from the feature point p. T is the ratio threshold, and this embodiment uses T = 0.8.

[0083] In this embodiment, when the improved FLANN algorithm is used for feature point matching, as long as the video screen is not moving at high speed, the algorithm can accurately find the matching points corresponding to the feature points.

[0084] In this embodiment, the specific method of performing natural language processing on the collected text data includes:

[0085] Perform text cleaning on text data, remove special characters, stop words, and convert uppercase and lowercase letters in the text;

[0086] Perform word segmentation on the text to divide the text into multiple independent words;

[0087] Mark the parts of speech of words.

[0088] In this embodiment, the specific method of using the SVM classifier to construct the literacy evaluation model includes:

[0089] Determine student ability indicators based on students' experimental operation standards;

[0090] Collect students' data on ability indicators, organize and process them, and divide the processed data into training sets and test sets;

[0091] The training set is used to train the SVM classifier, and the test set is used to input the SVM classifier for testing to obtain the literacy evaluation model.

[0092] The Support Vector Machine (SVM) algorithm is a data mining method based on statistical principles, which can effectively solve classification problems. The core idea of ​​the algorithm is to find an optimal classification hyperplane to maximize the interval area on both sides of the hyperplane, while ensuring the accuracy of classification. The SVM algorithm has no local minimum problem and does not need to rely on overall data. At the same time, the SVM algorithm can also solve high-dimensional feature space problems, and its generalization ability is strong. The literacy evaluation model constructed using the SVM classifier selects appropriate ability indicators and evaluation criteria based on the actual situation of students' experimental operations, which can accurately reflect students' experimental operation ability and report writing ability.

[0093] Through experimental activities, students can not only raise questions, design experiments, collect data, analyze results, summarize and reflect, but also develop critical thinking, innovative consciousness and teamwork ability. The cultivation of such abilities is not only an important goal of education, but also a basic quality required for social development. Among them, raising questions is the curiosity of students, and the level of curiosity varies depending on the behavior of curiosity. For example: often asking simple questions such as "What is this?" and "Why is it like this?"; liking to watch and imitate the behavior of adults or peers, trying to learn new skills, etc., is the performance of the first level; starting to ask more in-depth questions such as "Why is it like this?" and "How did you do it?", seeking detailed explanations, learning to find answers to questions of interest independently through resources such as books and the Internet, etc., is the performance of the second level of curiosity; students can carefully observe, think and analyze, and then raise complex, multi-dimensional, and worthy questions; around the questions they want to explore, they carry out visits, experiments and other research actions, and in order to explore the results, they are not afraid of frustration and have patience, etc., which is the performance of the third level of curiosity. Through the students' behavior during the experimental operation, we can get the students' research awareness, research ability and research character. Research awareness includes curiosity, imagination, and the ability to seek knowledge. Research ability includes the ability to find problems, solve problems, summarize and reflect. Research character includes the ability to seek truth, collaborate autonomously, and take responsibility. Based on the students' curiosity, imagination, curiosity, research awareness, research ability, research character, finding problems, solving problems, summarizing and reflecting, and report writing, the evaluation levels are divided into primary, secondary, and tertiary levels. Collect these indicators and the corresponding evaluation values ​​and corresponding levels of students during the experiment, sort them out and process them, divide the processed data into training sets and test sets, use the training set evaluation values ​​and corresponding levels as the input data of the SVM classifier to train the SVM classifier, and use the test set to test the performance of the SVM classifier, and finally obtain the literacy evaluation model. When evaluating students' research literacy ability, you only need to input the corresponding evaluation value into the SVM evaluation model to automatically output the evaluation result, and the corresponding result is the evaluation level of research literacy.

[0094] The embodiment of the present invention provides a method for tracking and dynamically evaluating the research literacy of primary school students throughout the process. By collecting various data of students in research activities, preprocessing the collected image data, performing natural language processing on text data, and using the constructed SVM classifier to construct a literacy evaluation model, the experimental operation behavior and information obtained after processing are quantitatively analyzed to obtain evaluation data of students' experimental operation ability and report writing ability, that is, students' research literacy evaluation data, and then sorting the evaluation data according to timestamps to construct a time series data set, using the time series analysis method to analyze the trend of students' learning performance, discovering students' progress and deficiencies, and finally displaying the evaluation data and learning performance trends in the form of charts. The method automatically collects and processes various types of data, realizes a comprehensive, accurate and objective evaluation of students' research literacy, and can dynamically track students' performance in real time.

[0095] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. An intelligent system for tracking and dynamically evaluating primary school students' research literacy, characterized by: include: Acquisition module, preprocessing module, feature matching module, text processing module, model processing module, analysis module and graphic display module; The acquisition module is used to collect image data, experimental operation image data, experimental report text data and discussion and communication text data of students in their research activities; The preprocessing module is used to preprocess the collected image data to obtain a preprocessed image; The feature matching module is used to extract features from the preprocessed image using the SIFI algorithm, and to perform feature matching using the improved FLANN algorithm to identify the students' experimental operation behaviors; The text processing module is used to perform natural language processing on the collected text data to obtain processed information; The model processing module is used to use the literacy evaluation model constructed by the SVM classifier to perform quantitative analysis on the experimental operation behavior and the processed information, and obtain evaluation data of the students' experimental operation ability and report writing ability; The analysis module is used to sort the evaluation data according to the timestamp, construct a time series data set, and analyze the time series data set using a time series analysis method to obtain the learning performance trend of the students; The graphic display module displays the evaluation data and learning performance trends in the form of charts.

2. The system according to claim 1, characterized in that The feature matching module includes a feature extraction unit, and the feature extraction unit is used to obtain a Gaussian difference scale space by convolving Gaussian difference kernels of different scales with the image; Detect the extreme points in the Gaussian difference scale space; Using the sub-pixel interpolation method, the extreme points and the sub-pixel positions of the extreme points in the continuous space are obtained by continuously interpolating the discrete spatial points, and the positions and scales of the key points are determined; Remove irrelevant feature points from all detected feature points; Determine the direction of the feature points and generate SIFT feature descriptors.

3. The system according to claim 2, characterized in that: The feature matching module further includes a matching unit, which uses a RANSAC algorithm to find a mapping relationship between the matching points based on the matching point pairs obtained from the previous frame image to obtain a mapping matrix; Search within the mapping point area of ​​the previous frame matching feature point to determine whether a matching point with a minimum Euclidean distance less than the set value is found. If so, match it; If not, determine whether the trace signs of the feature points are the same; If they are the same, the original FLANN algorithm is used for matching; If they are not the same, no match is performed.

4. The system according to claim 1, characterized in that The text processing module includes a text cleaning unit, a word segmentation unit and a tagging unit, wherein the text cleaning unit is used to perform text cleaning on the text data, remove special characters, remove stop words, and convert the letters in the text into uppercase and lowercase letters; The word segmentation unit performs a word segmentation operation on the text to divide the text into multiple independent words; The marking unit marks the part of speech of the word.

5. A method for full-process tracking and dynamic evaluation of primary school students' research literacy, characterized in that: include: Collect image data of students' research activities, experimental operation image data, experimental report text data and discussion and communication text data; Preprocessing the collected image data to obtain a preprocessed image; The SIFI algorithm is used to extract features from the preprocessed images, and the improved FLANN algorithm is used for feature matching to identify students' experimental operation behaviors. Perform natural language processing on the collected text data to obtain processed information; The literacy evaluation model constructed by SVM classifier is used to quantitatively analyze the experimental operation behavior and processed information to obtain the evaluation data of students' experimental operation ability and report writing ability; Sort the evaluation data according to timestamps, construct a time series data set, and use time series analysis methods to analyze the time series data set to obtain the students' learning performance trends; Present assessment data and learning performance trends in charts and graphs.

6. The method according to claim 5, characterized in that The specific method for preprocessing the collected image data includes: image denoising, image enhancement, image geometric transformation and image normalization processing.

7. The method according to claim 5, characterized in that The specific method of extracting features from the preprocessed image using the SIFI algorithm includes: The Gaussian difference scale space is obtained by convolving the image with Gaussian difference kernels of different scales; Detect the extreme points in the Gaussian difference scale space; Using the sub-pixel interpolation method, the extreme points and the sub-pixel positions of the extreme points in the continuous space are obtained by continuously interpolating the discrete spatial points, and the positions and scales of the key points are determined; Remove irrelevant feature points from all detected feature points; Determine the direction of the feature points and generate SIFT feature descriptors.

8. The method according to claim 7, characterized in that The specific method of performing feature matching using the improved FLANN algorithm includes: According to the matching point pairs obtained in the previous frame image, the mapping relationship between the matching points is obtained using the RANSAC algorithm to obtain the mapping matrix; Search within the mapping point area of ​​the previous frame matching feature point to determine whether a matching point with a minimum Euclidean distance less than the set value is found. If so, match it; If not, determine whether the trace signs of the feature points are the same; If they are the same, the original FLANN algorithm is used for matching; If they are not the same, no match is performed.

9. The method according to claim 5, characterized in that The specific method of performing natural language processing on the collected text data includes: Perform text cleaning on text data, remove special characters, stop words, and convert uppercase and lowercase letters in the text; Perform word segmentation on the text to divide the text into multiple independent words; Mark the parts of speech of words.

10. The method according to claim 5, characterized in that The specific method of using the SVM classifier to construct the literacy evaluation model includes: Determine student ability indicators based on students' experimental operation standards; Collect students' data on ability indicators, organize and process them, and divide the processed data into training sets and test sets; The training set is used to train the SVM classifier, and the test set is used to input the SVM classifier for testing to obtain an evaluation model.

Images