Evaluation of Teaching Reform Effect of Computer Vision Specialty

Based on Deep Learning

Yajing Chen

Quanzhou University of Information Engineering, Quanzhou, China

Keywords: Deep Learning, Computer Vision, Educational Reform, Specialty, Effect, Assessment.

Abstract: Most of the conventional evaluation methods for teaching reform effect are designed based on the principle

of K-modes algorithm, with limited scope of application, large deviation of evaluation results, and unable to

obtain more accurate teaching reform effect. This paper introduces the principle of deep learning method,

and takes the computer vision specialty as an example, puts forward the evaluation research on the teaching

reform effect of computer vision specialty based on deep learning. First, select the data set required for

evaluation, and preprocess the data set to reduce the repeatability and redundancy of data. Secondly,

establish the evaluation index system of the teaching reform effect of computer vision specialty, analyze and

find out the potential internal links in the data, as the theoretical basis for the evaluation of the teaching

reform quality effect. On this basis, hierarchical nesting is used to characterize the affiliation of deep

learning and evaluate the effect of teaching reform. It can be seen from the application test results that after

the application of the new evaluation method, the average score evaluation result of the computer culture

basic application skills test is closer to the actual score, with small deviation and significant advantages in

evaluation effect.

INTRODUCTION

Reforming education and teaching has a significant

impact on enhancing the quality of talent

development. As a crucial focus of higher education,

it serves as a powerful driving force for developing

education and teaching in the current era (Ni Z,

2022).In a narrow sense, it includes the reform of

teaching methods, teaching models, teaching means

and other aspects(Steinberg, 2021).With the

development of university reform and the deepening

of the people, emphasizing the value orientation of

people centered management is to focus on the

student groups that play a decisive role in improving

the quality of education, which helps to improve the

quality and efficiency of university

management(Armstrong, 2021). According to the

development of higher education in China, the

undergraduate teaching reform in local colleges and

universities has always been the focus of attention of

the whole society and the Ministry of Education

(Gao H. Reform, 2021). In recent years, the scale of

higher education in China has been constantly

expanding, with many universities and majors. This

rapid expansion of scale and content has indeed met

the expectations of society, enterprises and every

family for higher education, accelerated the speed of

talent cultivation to a certain extent, and made up the

demand gap of social and economic development for

college graduates(Li W . Role, 2021).

According to the current situation of

undergraduate teaching reform in local colleges and

universities, many colleges and universities only

stay at the superficial level, often focusing on the

content and methods, and ignoring the "teaching

structure"(Chen X. Study, 2022). That is, under the

guidance of specific educational concepts and

theories, a stable teaching and research model (Hui

Y. Evaluation, 2021) suitable for specific teaching

environment has gradually formed. Deep learning

refers to the process in which learners actively

participate in learning activities, critically learn new

knowledge, and integrate new knowledge into the

existing knowledge structure under the guidance of

understanding (Feng W., 2022). It shows that

learners have reached the level of knowledge

transfer on the basis of understanding, and can

effectively solve the complex problems in the

evaluation of teaching reform effect under the new

situation(Wu X ., 2022).Deep learning focuses on

understanding, emphasizes transfer, can take their

knowledge and experience as a kind of thinking

208

Chen, Y.

Evaluation of Teaching Reform Effect of Computer Vision Specialty Based on Deep Learning.

DOI: 10.5220/0012277700003807

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 2nd International Seminar on Artiﬁcial Intelligence, Networking and Information Technology (ANIT 2023), pages 208-213

ISBN: 978-989-758-677-4

activity, better understand the connection and

essence of things, and will learn professional

knowledge and skills for use in the relevant

environment (Ensminger, 2021).

When higher education institutions implement

reforms, students are one of the primary groups

evaluated for the effectiveness of the reforms.

However, due to the modifications in teaching

content, methods, and goals, evaluating the

effectiveness of teaching reforms faces certain

challenges (Liu, 2022).In order to improve the

applicability and feasibility reform effect, this paper

introduces the principle of deep learning method,

selects computer vision specialty as the research

goal, and puts forward the research on the evaluation

of teaching reform effect of computer vision

specialty based on deep learning, which provides

reference suggestions for promoting the smooth

progress of various education and teaching reforms

in local colleges and universities.

EVALUATION OF TEACHING

REFORM EFFECT OF

COMPUTER VISION

SPECIALTY BASED ON DEEP

LEARNING

2.1 Data Set Selection and

Pre-Processing

In the evaluation method of teaching reform effect

of computer vision specialty based on deep learning

designed in this paper, firstly, the data set needed for

this evaluation needs to be selected and

preprocessed. After the data is captured, the

corresponding database is established, and the

acquired data is injected into the database

[

Guo J ,

2021

]

.According to the design requirements, eight

tables are set in the database, which are: teacher

information table, teacher education table, teacher

professional title table, teacher timetable, student

information table, student curriculum table, teacher

evaluation table and student employment table

(

Zhang, 2021).The data will lay a solid foundation for

future data mining work (

Li, 2021).

To realize data integration, data integration is

actually data consolidation. In the process of

consolidation, heterogeneous data can be

consolidated and disposed. The general processing

methods include entity identification, redundancy

processing and data conflict processing (

Cahyadi,

2021). The specific steps and descriptions are shown

in Table 1.

Table 1: Heterogeneous data processing methods and steps

for teaching reform of computer vision specialty.

Processing

metho

Explain Step

Dataset

entity

recognition

It refers to the situation where

an entity corresponds to

multiple data sources.

Implement entity

differentiation and

lassification through patter

integration of the initial data

rom the experimental objec

database or data warehouse.

Dataset

redundancy

Refers to the phenomenon of

multiple occurrences of data

with the same attributes in the

selected evaluation data.

Use filtering algorithms to

filter out redundant parts

within the dataset.

Dataset

detection

and

processing

of data value

conflicts

It refers to the phenomenon

that when there is an entity

corresponding to multiple data

sources in the evaluation

dataset, there are multiple

different attributes between

the data, resulting in conflicts

etween the data.

Implement data analysis

using redundancy detection

algorithms to eliminate

duplicate parts.

The data set is integrated and processed

according to the heterogeneous data processing

method shown in Table 1. The necessity of data

integration is mainly reflected in two aspects: one is

to stagger the composition errors of data in the

framework; The second is to avoid and reduce the

duplication and redundancy of data. Data integration

enables more physical storage space for data (Zhang

J ., 2021).

2.2 Establish the Evaluation Index

System of Teaching Reform Effect

After completing the selection and pre-processing of

the evaluation data set of the teaching reform effect,

the next step is to establish the evaluation index

system of the teaching reform effect of computer

vision specialty, which provides an important

reference basis for the subsequent evaluation of the

teaching reform effect. First of all, in the indicator

system, a computer vision professional teacher

information table is established to multi-dimensional

reflect the relevant data information of teacher

quality indicators, as shown in Table 2.

Table 2: Information of Computer Vision Teachers.

Field Name Field Descri

tion T

e Len

JS-GH Teacher ID Int 8

JS-MM Passwor

Varcha

JS-XM Full name Varcha

JS-XL

The highest education

level

Varchar 20

JS-ZC Title Varcha

JS-GL Workin

ears Datetime 4

Evaluation of Teaching Reform Effect of Computer Vision Specialty Based on Deep Learning

209

As shown in Table 2, there are certain differences in

the attribute fields corresponding to each attribute

item.The evaluation information of the teaching

reform effect comes from each semester, and the

students evaluate and grade the teachers of the

specialty through the school system, and generate

the evaluation information table of the teaching

reform effect, as shown in Table 3.

Table 3: Evaluation Information of Teaching Reform

Effect.

Field Name Field Descri

tion T

e Len

PJ-ID Course ID Int 8

PJ-GH Teacher ID Varchar 8

PJ-JS Course acceptance Varchar 20

PJ-HD Course interactivity Varchar 20

PJ-YY Language expression ability Varchar 20

PJ-CY Student engagement Varchar 20

PJ-KH Assessment method Varchar 20

PJ-ZH Comprehensive evaluation Varchar 20

Through the evaluation information table of teaching

reform effect of computer vision specialty in Table

3, analyze and find out the potential internal

relations in the data as the theoretical basis for the

evaluation of teaching reform quality effect.

2.3 Evaluation of Teaching Reform

Effect Based on Deep Learning

The next step following the establishment of the

teaching reform evaluation index system is to utilize

a comprehensive and multi-dimensional in-depth

learning approach to evaluate the effectiveness of

computer vision teaching reform. Firstly, an analysis

of the membership relationships within deep

learning will be conducted, with hierarchical nesting

used to characterize these relationships, as depicted

in Figure 1.

Figure 1: Schematic diagram of deep learning membership

relationship.

As shown in Figure 1, the depth of the deep

learning model does not have a standard value, and

its abstract features are calculated from relatively

less abstract features (Hong, 2021). On this basis, a

teaching reform effect evaluation team led by the

president and members of the school staff was

established to make an objective and correct

assessment of the teaching reform effect of all

teachers (Cui, 2022). According to the mid-term or

final examination results, the teachers will conduct

two evaluations every school year, sometimes a

person by person contest (Wang, 2021). The average

value of the two evaluation results is the final

evaluation result of the school year (the average

value of four evaluations can be used in junior high

school) (Chamorro-Atalaya O, 2021). The central

school will report the assessment results (Chen Y.

Evaluation, 2021) to the school and teachers

themselves in the form of briefing. The evaluation

process of teaching reform effect of computer vision

specialty based on deep learning designed in this

paper is shown in Figure 2.

Figure 2. Evaluation process of teaching reform effect of

computer vision specialty based on deep learning.

ANIT 2023 - The International Seminar on Artiﬁcial Intelligence, Networking and Information Technology

210

As shown in Figure 2, after the establishment of the

assessment team, the evaluation data value of the

teaching reform effect is converted, and each

different sub type data is assigned a rank (

Zhang,

2022

) that matches its value. Set the transformation

function of the evaluation value of teaching reform

effect as 𝑓



𝑥



, the expression is:

(1)

Among them,

represents the evaluation value

of the original teaching reform effect.Secondly,

calculate the similarity of evaluation frequency of

teaching reform sample data, and the formula is:

𝐴𝑉𝐹



𝑥











∑





𝑓𝑥



 (2)

Among them,

represent the sample data of

teaching reform effect evaluation of computer vision

specialty;

represents the sample data

dimension; 𝑓𝑥



indicates that the sample data is in

the attribute

frequency on (Xie, 2021).Through

calculation, the similarity of sample data evaluation

frequency is obtained, and the similarity is used to

judge the relationship between sample data volume

and sample data dimensions, and then test whether

there is deviation in the evaluation of teaching

reform effect (Luo, 2021). If the evaluation

frequency of sample data is too similar, it means that

there is a large deviation in the assessment results,

and vice versa.Through the above steps, the goal of

the reform effect assessment of computer vision

specialty (Wang, 2022) can be achieved.

APPLICATION TEST

3.1 Test Preparation

The above content is the whole design process of the

evaluation method for the teaching reform effect of

computer vision specialty proposed in this paper by

using the principle of deep learning method.Before

putting forward the evaluation method for use in the

actual teaching reform work, it is necessary to make

an objective test of the feasibility and evaluation

effect of the method to ensure the effectiveness of its

evaluation results and avoid losses caused by direct

use.Through the "multi-stage stratified sampling"

method, that is, first determine the survey area, then

conduct random cluster sampling with the teaching

class as a unit, and finally determine the survey

students.Students from Computer Science College of

S University were selected as the research goal, and

some students from 2020 and 2021 were randomly

selected for the wake up test to simulate the

computer vision professional skills test.In the

process of learning assessment, only 52% of the

schools carried out learning assessment strictly in

accordance with the five areas, and 48% of the

schools did not use the five areas of the new

curriculum standards to evaluate.The reason is that

9% disapprove of the new curriculum assessment,

22% think the assessment system is too complex and

the workload is too heavy, 19% think the assessment

is subjective, and 60% do not have the objective

conditions of the school.In fact, even though

teachers are highly recognized subjectively, the

objective conditions of the school also limit the

establishment and improvement of the new

evaluation system.In order to describe the effect of

online teaching from the perspective of students'

in-depth learning, this paper gives a simple total

score for each dimension, and then obtains its

average value to compare each dimension more

accurately and effectively.

Table 4: Scoring Standards for Application Skills Test of

Computer Basic Courses.

Question numbe

Sam

lin

oint fraction

Windows operation

uestions

(

oints

)

Can create a folder named after the

student ID.

Able to establish 3 correctly named

subfolders of the same level.

Word operation

uestions

(

oints

)

Insert text material into document 1. 2

Set fonts, paragraphs, shading,

orders, table row height, table

column width, and drop caps as

uired.

Add document naming, title, table file

namin

Excel operation (20

points)

Create an Excel workbook with the

correct name and rename the

worksheet correctl

Able to in

ut worksheet content. 5

The calculation of growth rate and

evaluation

rowth rate is correct.

Powerpoint operation

uestions

(

oints

)

Can set slide layout, font, title,

ack

round, and switchin

methods.

The presentation is named and stored

in the correct location.

Can complete animation settings as

uired.

Network operation

uestions

(

oints

)

Can summarize according to the

search information and re

uirements.

Able to write the content of the letter

as required.

Can compress completed files and

send them to the s

ecified email.

Evaluation of Teaching Reform Effect of Computer Vision Specialty Based on Deep Learning

211

3.2 Test Results

Set the evaluation method of computer vision

teaching reform effect based on deep learning

proposed in this paper as the experimental group,

and set the traditional evaluation method of teaching

reform effect as the control group, obtain the

evaluation results of computer vision teaching

reform effect of the two methods, and compare them

to make the application test results more convincing.

First of all, select the basic course of computer

vision application skills course as the basis for this

application test, and set the scoring criteria for the

basic course of computer application skills test, as

shown in Table 4.

As shown in Table 4, it is the scoring standard of

computer vision teaching reform for this application

test.145 students participated in the computer culture

basic application skills test. Among them, 72 are

from Grade 20 and 73 are from Grade 21;There are

55 boys and 90 girls.The average scores of science,

arts, sports and art students are 87, 85, 84 and 79

respectively.The Windows operation question, Word

operation question, Excel operation question,

Powerpoint operation question and network

operation question are labeled with R1~R5

respectively.Use the above two methods to evaluate

the average score of computer culture basic

application skills test, and compare it with the actual

average score. The results are shown in Figure 3.

Figure 3: Evaluation Results of Computer Culture Basic

Application Skills Test.

It can be seen from Figure 3 that after the

application of the two methods, there are obvious

differences in the results of performance

evaluation.Among them, after the application of the

assessment method proposed in this paper, the

average score evaluation result of the computer

culture basic application skills test is more close to

the actual score with less deviation, which indicates

that the assessment method proposed in this paper

has higher accuracy, more accurate evaluation

results and higher feasibility.

CONCLUSION

An effective and rational method for evaluating the

impact of teaching reforms has a significant

influence on enhancing the standard of education

and pedagogy. In conclusion, to address the

limitations and deficiencies of conventional

evaluation methods employed in professional

education and teaching reforms, a more

comprehensive approach must be implemented in

practice, this paper introduces the principle of deep

learning methods, and proposes an all-round

research on the computer vision professional

teaching reform effect evaluation based on deep

learning.Through the research in this paper, the

accuracy of the evaluation results of professional

teaching reform effect has been effectively

improved, and the teaching reform effect of

computer vision specialty has been mastered from a

deeper level, which provides an important guarantee

for talent cultivation, and has important research

significance in helping higher institution promote

the innovative development of education and

teaching.

REFERENCES

Ni Z, Wang F. Quality Assessment of Vocational

Education Teaching Reform Based on Deep

Learning[J]. Computational and mathematical methods

in medicine, 2022, 2022:1499420. DOI:

10.1155/2022/1499420

Steinberg M, Bourgeois I, Wildschut L . How does

teaching with cases support the development of

evaluation competencies? [J]. New Directions for

Evaluation, 2021, 2021(172):37-51. DOI:10.1002/EV.

20486

Armstrong S N, Lupinski K, Burcin M M, et al.

Evaluation of a Teaching Assistant Program in Online

Education[J]. Journal of Educational Research and

Practice, 2021, 11(1). DOI: 10.5590/JERAP.2021.

11.1.04.

Gao H . Reform of College English Teaching Model under

the Background of Artificial Intelligence[J]. Journal of

Physics: Conference Series, 2021, 1744(4):042161

(6pp). DOI: 10.1088/1742-6596/1744/4/042161

Li W . Role of machine learning and artificial intelligence

algorithms for teaching reform of linguistics[J].

Journal of intelligent & fuzzy systems: Applications in

ANIT 2023 - The International Seminar on Artiﬁcial Intelligence, Networking and Information Technology

212

Engineering and Technology, 2021, 40(2):3251-3262.

DOI: 10.3233/JIFS-189365

Chen X. Study on Teaching Reform Based on the

Investigation of College Students'Entrepreneurial

Environment[J]. Asian Agricultural Research, 2022,

14(6):60-63. DOI: 10.22004/ag.econ.333623

Hui Y. Evaluation of Blended Oral English Teaching

Based on the Mixed Model of SPOC and Deep

Learning[J]. Scientific programming, 2021(Pt.10):2021.

DOI: 10.1155/2021/7044779

Feng W. Wireless Sensor Technology-Based Physical

Education Teaching Evaluation[J]. Computational

intelligence and neuroscience, 2022, 2022:3544457.

DOI: 10.1155/2022/3544457

Wu X. Research on the Reform of Ideological and

Political Teaching Evaluation Method of College

English Course Based on "Online and Offline"

Teaching[J]. Journal of Higher Education Research,

2022, 3(1):87-90. DOI: 10.32629/jher.v3i1.641.

Ensminger D C, B Montrosse‐Moorhead, Frazier E W ,

et al. How do we deepen our story reservoir by

designing, developing, and writing instructional cases

for teaching evaluation? [J]. New Directions for

Evaluation, 2021, 2021(172):85-102. DOI:

10.1002/EV.20484

Liu G, Zhuang H . Evaluation model of multimedia-aided

teaching effect of physical education course based on

random forest algorithm[J]. Journal of Intelligent

Systems, 2022, 31(1):555-567. DOI: 10.1515/jisys

-2022-0041.

Guo J, Bai L, Yu Z , et al. An AI-Application-Oriented

In-Class Teaching Evaluation Model by Using

Statistical Modeling and Ensemble Learning[J].

Sensors, 2021, 21(1):241. DOI: 10.3390/s21010241.

Zhang C , Guo Y .Mountain rainfall estimation and online

English teaching evaluation based on RBF neural

network[J].Arabian Journal of Geosciences, 2021,

14(17):1-11. DOI: 10.1007/s12517-021-08038-7.

Li W . Li W .Urban land ecological evaluation and English

distance teaching evaluation based on wireless

sensors[J].Arabian Journal of Geosciences, 2021,

14(17):1-15. DOI: 10.1007/s12517-021-08162-4.

Cahyadi A , Hendryadi, Widyastuti S , et al. Emergency

Remote Teaching Evaluation of the Higher Education

in Indonesia[J]. Heliyon, 2021, 7(8):e07788. DOI:

10.1016/J.HELIYON.2021.E07788

Zhang J . Research on Classroom Teaching Evaluation and

Instruction System Based on GIS Mobile Terminal[J].

Mobile Information Systems, 2021, 2021(11):1-11. DOI:

10.1155/2021/9790766.

Hong D, Gao Z, Luo J, et al. Emotion analysis of

teaching evaluation system based on AI techno-logy

towards Chinese texts[J]. MATEC Web of Conferences,

2021, 336:05007. DOI: 10.1051/matecconf/202133

605007.

Cui Y, Han G, Zhu H. A Novel Online Teaching Effect

Evaluation Model Based on Visual Question

Answering[J]. Journal of Internet Technology, 2022,

23(1):91-98. DOI: 10.53106/160792642022012301009.

Wang L, Rong L, Teng F, et al. Teaching Performance

Evaluation Based on the Proportional Hesitant Fuzzy

Linguistic Prioritized Choquet Aggregation Operator[J].

International Journal of Computational Intelligence

Systems

, 2021, 14(1):635-650. DOI: 10.2991/ijcis.d.

210112.001.

Chamorro-Atalaya O, Morales-Romero G, Trinidad-Loli

N, et al. Evaluation of Teaching Performance in the

Virtual Teaching-Learning Environment, from the

Perspective of the Students of the Professional School

of Mechanical Engineering[J]. International Journal of

Emerging Technologies in Learning (iJET), 2021,

16(15):244-252. DOI: 10.3991/ijet.v16i15.23091.

Chen Y . Evaluation of Teaching Effect of Internet of

Things Education Platform based on Long-term and

Short-term Memory Network[J]. International Journal

of Continuing Engineering Education and Life-Long

Learning, 2021, 31(1):36-52. DOI: 10.1504/IJ

CEELL.2021.10028912.

Zhang Y , Yang Y . The evaluation method for distance

learning engagement of college English under the

mixed teaching mode[J]. International Journal of

Continuing Engineering Education and Life-Long

Learning, 2022, 32(2):159-175. DOI: 10.1504/IJ

CEELL.2022.121947

Xie Z, Su Z. Evaluation of college English classroom

teaching quality dependent on triangular fuzzy

number[J]. International Journal of Electrical

Engineering Education, 2021:002072092110020. DOI:

10.1177/00207209211002075.

Luo R , Liao Q , Zheng J , et al. A study on the

evaluation method of foreign language classroom

teaching process from the perspective of

neuroeducation[J]. International Journal of Electrical

Engineering Education, 2021:002072092098468. DOI:

10.1177/0020720920984686.

Wang H, Wang Z. Design of Simulation Teaching System

Based on Modular Production and Processing[J].

Computer Simulation, 2022, 39(4):205-209. DOI:

10.3969/j.issn.1006-9348.2022.04.041.

Evaluation of Teaching Reform Effect of Computer Vision Specialty Based on Deep Learning

213