The Application of Artificial Intelligence Technology in Art Teaching Taking Architectural Painting as an Example.

In the current era of technology, artificial intelligence has grown rapidly in such a way that it has established its presence in all fields. The purpose of artificial intelligence is to reduce human intervention and complete tasks with an enhanced result. In this research, we are going to study the application of artificial intelligence technology in art teaching, taking architectural painting as an example. Architectural painting is a type of painting that focuses only on architecture, including indoor and outdoor views of the buildings. In earlier stages, architecture was shown only in the background of paintings that had different objects as the main subject. Later, architecture itself became a mainstream genre in the field of painting. As has been shown by other researchers, the latest technologies such as Internet technology, wireless sensor networks (WSNs), and artificial intelligence like deep learning technologies are deployed in art teaching. Artificial intelligence has made teaching easier. This proposed system makes use of Internet technology, WSNs, artificial intelligence, and lightweight deep learning models in the field of art teaching. The teaching method is enhanced by adapting to this new technology. For performing the analysis of the proposed system, the Limited Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) art algorithm is implemented. This L-BFGS algorithm focuses on finding the local minima in any given application. In this art teaching of architectural painting, the proposed algorithm will aid in explaining the minute works to be noted while doing the artwork. The proposed algorithm is then compared with the traditional Gradient Descent, Adam, and Adadelta algorithms. From the results, it can be observed that the proposed algorithm has achieved accuracy of 97% and 98% in the training and testing phases, respectively.

1. Introduction

Digital museums, digital libraries, and associated websites for Chinese painting and art have emerged in the age of the Internet and the World Wide Web [1]. Traditional Chinese paintings are becoming more popular because of the ease with which they may be preserved and accessed. It is also straightforward to digitise the Xuan paper Chinese paintings and create digital pictures of the paintings in the traditional style. Because of this, digitising Chinese art is gradually becoming a new way to protect it. Specialists are unable to quickly identify the kind of painter when presented with unsigned digital images of Chinese art in diverse styles, which has a negative impact on the efficiency of authenticating work [2]. Thus, as an essential area for the digital Chinese painting authenticity-aided verification process, it has become more vital to research the accuracy and automation of categorisation of Chinese painting artists. However, the widespread digitisation of Chinese paintings has made it hard to check the quality of digital Chinese paintings in databases [3]. To make matters worse, as the use of big data has increased, the legitimacy of digital Chinese artworks has become a growing cause for worry. When it comes to traditional identification techniques, professionals' extensive knowledge is the most important asset. Forgeries using modern information technology pose a challenge to existing techniques of authentication because they lack objective referenceable signs and are prone to human bias [4]. An automated and accurate digital Chinese painting identification tool is needed to help authenticate and identify the works.

Inspired by Cubism, Futurism, and Dadaism art movements, digital interactive art is a continuation of these creative movements. People may go all the way back to 1956, when the term “artificial intelligence” was coined, to understand the connection between artificial intelligence and interactive art [5]. It has been clear from the beginning that interactive art and artificial intelligence will be intertwined because of the intrinsic relationship between computer technology and art. At the point when intuitive workmanship was at its outset, its makers centred around the feel of the medium's actual structure and its innovative execution, ignoring the client's insight and mental mode to fulfil the style of structure and application [6]. During the 1980s, PC brain science arose as a mental field, fake neural organisation research advanced quickly, and revelations in voice acknowledgment and machine vision advancements moved man-made reasoning examination into its outset [7]. During this time, the notion of experience was introduced to artists and technologists as a means of creating new forms of interactive art.

To ensure the long-term growth of art and HCI technology, it is essential that these two fields work together harmoniously to produce new ideas and new products [8]. Overall, artificial intelligence is much more than a basic technological tool for art production; it is a reshaping of art creation thought and an influence on human cognition [9]. Research in sociology, psychology, management, and film and television is used to provide a thorough theoretical foundation for this dissertation, which is supported by a wide range of scholarly works in the fields of film and television production and market management and the viewing experiences of actual moviegoers [10]. For a deeper understanding of the subject, the study uses both macro and micro analysis. When it comes to interactive art expression, artificial intelligence (AI) deconstructs and recreates the original model in terms of technology and thinking; then, it analyses the influence of AI on interactive art expression in the context of time and space [11]. It also establishes its application paradigm. Based on these results, there may be a new practical approach to interactive art. A wide range of external environmental elements may be analysed using correct statistical methodologies to understand the variations in shape and quantity of distribution, the relationship between variables, and more. Because of its obvious indirectness, this research technique is less disruptive and easier to implement [12]. In order to obtain a sense of the present level of AI development in the film and television business, one may look at the status of AI firms, products, and technical advances that are relevant to the film and television industry and the use of AI in this sector.

Even before the term “artificial intelligence,” there was extensive global research on the subject [13]. Artificial intelligence (AI) has made significant strides in recent years, mostly due to the emergence of new technologies that build on its foundation. Since general society and useful applications of artificial reasoning (AI) have recently become more common, research in the field of entertainment is primarily centred on diary reports. Furthermore, the examination content focuses on AI as a film content component and on the effect of AI on creation, market board, and buyer experience in the entertainment world chain. The properties of Chinese fine arts are utilised as the establishment for keen order, and the amount of data they contain straightforwardly impacts grouping results [14]. Wavelet traits are used to classify Chinese artworks as one of many characteristics. A categorisation system based on information theory is suggested for paintings based on feature values that reflect their properties. Studies of Chinese painting writers take advantage of the paintings' regional and global characteristics. After further investigation, the creative style of the painter becomes an essential aspect in the identification of the picture. Painting characteristics such as colour, texture, and brushstroke all play an increasingly significant role in the process of identifying different types of artists. Images of the target may be classified using their colour and form [15]. There are two categories based on brushwork and painting that use colour histograms, colour coherence vectors, and autocorrelation texture features [16]. For classification purposes, the colour attributes of brushstrokes and brushwork in Chinese artworks are retrieved. It is now possible to fully classify Chinese artworks using a novel method called multiscale grayscale covariance matrix extraction. By using colour, texture, and brushstrokes, it is possible to better convey the painting's creative style and boost classification accuracy. However, the essential underlying qualities lead directly to bad classification results and have constraints [17]. They rely on the visual substance of the images. Chinese paintings' colour and form traits may serve as a basis for future research into multidimensional characteristics based on image content, even if such studies have not yet existed. A lack of underlying features in studies like those cited above makes it difficult to do proper analysis.

In addition to the validity of the features themselves [18], the classification technique and its handling of characteristics are also critical to the success of intelligent classification. Convex packet models are used to categorise Chinese artworks. It is difficult to automate categorisation using the statistical models outlined above because of the time and effort required to adjust parameters. They used support vector machines to classify Chinese paintings after extracting wavelet characteristics from the artworks they were working with. Machine learning-based techniques improved parameter adjustment, but feature quantification remained unsatisfactory. According to the study, the end-to-end architecture of deep learning-based classification algorithms not only boosts the automation of the classification process, but also provides a high-level semantic representation of picture information. Picasso's brushstroke properties are analysed using recurrent neural networks (RNNs) to classify the artist's work [19]. In order to categorise and explain the texture features of ancient artworks, deep convolutional neural networks (DCNNs) are used. A CNN is used to extract brushstroke features from ink paintings in order to quantify and identify the style of the artwork. For the CNN model to better recognise patterns, an aggregate structure with several layers is used. Jump connections and residual mechanisms are proposed to increase the depth of the network in order to improve classification performance using residual processes. A feature rescaling module is being investigated for inclusion in order to call attention to the relevance of distinct image feature channels. In order to enhance the extraction of features from CNN models, the convolutional block attention mechanism module (CBAM) is being developed. The use of improvement modules to increase the learning capacity of deep learning networks has therefore become a key trend in intelligent categorisation.

Only a few of the new creative conceptions investigated in the Western art sector in the 1960s and 1970s included theories on artificial intelligence and interactive art, such as artificial life art and “cyber art” [20]. In the beginning, the focus was on artists working at the bleeding edge of scientific discovery and technological advancement. Scientific subjects and technological categories are broken down into sections in the book, which addresses how art, science, and technology, such as robots, interact and investigates new methods of exposing modern artists' work. Human and machine intelligence are predicted to merge during the technological singularity—enhancing the human brain with faster, more accurate, larger storage; stronger memory; and more efficient data-sharing capabilities [21]. When it comes to art, it is important to consider how artificial and technical systems will interact and how that will affect future intelligence's ability to inspire new forms of creative expression.

1.1. Motivation for the Study

Artificial intelligence (AI) is not frequently used in art education or architectural painting. As a consequence, the project's focus is to create techniques for introducing AI into art instruction. For that purpose, the authors examined the current status of AI applications for art instruction and summarised their shortcomings. Following that, the oversimplified role of AI in art teaching has been thoroughly examined, and techniques have been devised to inspire AI applications in the areas of modern art teaching in the following areas: broadening the flexibility of AI-based painting teaching, trying to improve the advanced education of art teaching, and improving the engaging creation and atmosphere of AI-based art teaching. Concurrently, a performing analytical framework was built to assess the impact of AI implementation in art instruction and architectural painting training. The purpose is to measure the influence of AI through art instruction and architectural painting arts programmes, which have a variety of applications.

2. Materials and Methods

The arrival of new technologies has made a great change in the creative industry. Computers are used in creative activities such as fine arts, music, and architecture. The computer itself plays the role of a musical instrument in the case of music; it also plays the role of brush in the case of painting. Beyond assisting humans in various tasks, the computer is a unique entity that can create art forms of its own. Artificial intelligence plays a major role in the creative industry. This is termed computational creativity. In this article, we are going to study how artificial intelligence and computer technology are used in art teaching especially in the case of architectural painting. Architectural painting is one of the traditional art forms. This type of painting is one among the various genres in the painting that especially concentrates on architectural buildings, including the indoor and outdoor views of those buildings. This type of art form emerged during the sixteenth century. The proposed model introduces an intelligent art teaching model which teaches art with the help of wireless sensor networks (WSNs) and Internet technology. In the intelligent teaching methodology, the classes are taken online. Students listen to their classes via smart devices such as mobiles, laptops, and personal data assistance. The classes are also taken with the help of apps and web pages. An intelligent art teaching model makes use of all the latest technologies to teach the art form. Here, architectural painting is taken as an example. The digital picture of the particular architecture is given to the students online along with the set of procedures that were carried out in creating a painting. The students are provided with a digital image of the architecture that has to be painted. With reference to the digital image of the architecture and with the help of instructions provided by the app, the students can create an architectural painting of their own with the help of AI technology. Prerecorded video classes are utilised for teaching. The advantage of prerecorded video classes is that they can be accessed any number of times, unlike the traditional classroom style. This helps the students to learn better until they clear their doubts. Nowadays, artificial technology has entered the art business by creating art through technology by integrating artists' pictures with predesigned templates to create an art form. This task is carried out with the help of apps, and this is termed the deep learning style transfer technique. When art teaching is considered, the art form has to be done manually as in the traditional method. However, the method of teaching is done online with the help of artificially intelligent technologies.

To summarise the proposed system (refer to Figure 1), the teaching of art, for example, how to paint an “architectural painting,” is taught via online classes with the help of prerecorded videos, interactive live classrooms, etc. This is accomplished with the help of devices such as mobiles and laptops with a wireless Internet connection at both ends. The digital photograph of the architecture is sent to the students for reference along with the set of principles to be followed as an instructional video. The students have to physically paint the architecture with water colours or oil paints as instructed. Since it is a creative art form, it depends on the skills of the students and how they explore the artwork. Every student has to complete the artwork and send it back to the art schools in digital format. This painting is validated by the teaching staff and taken for academic purposes. Thus, the teaching of art with the help of artificial intelligence is more efficient than the traditional way of art teaching.

Figure 1

Proposed model for art teaching.

To combat the declining influence on classical cultural signifiers, research on classical cultural symbols has become increasingly important. The goal of this study is to look into the use of classical cultural touchstones in art education and architectural painting direction in the context of artificial intelligence. The exploratory outcomes show that the improved Limited Broyden–Fletcher–Goldfarb–Shanno (L-BFGS) algorithm now has the lowest loss and also the shortest distance in much more than 500 s of time consumption. When compared to the conventional test, certain regular and reliable methods are declined by 65 percent. Its loss could be decreased by around 48 percent using the art approach. When contrasted to the newly improved Adam approach, its losses are reduced by about 9%, which is a substantial improvement. In this sense, an L-BFGS algorithm is basically a comprehensive recipe for creating and possibly implementing an artwork. This comprehensive recipe of the algorithms, functional areas, facial expressions, and some other input that ultimately determines the structure is included in the art teaching, but architectural painting is also considered as input. This input could be numerical, information-processing-related, or formative in nature.

To examine the application impact of AI through modern art education, the media downloads should be analysed. This study discussed experts and selected several identifiers to analyse the performance of the system of AI through art teaching and acquiring architectural painting teaching, including such art teaching methods, art teaching approaches, art teacher effectiveness, and teaching environment in acting classes. Using a professional teacher, a scoring method and a weighted judgments structure of performance indicators are applied and are constructed in the following manner.

AI performance measurement of modern art teaching is represented as in r00∗a+r01∗b+r02. In the traditional double function iteration, the sequential transformations model is applied. However, in a different perspective, r10∗a+r11∗b+r12  is being used to complete all nesting boxes for self-similarity. Using a 3 × 3 matrix, the interpolation transformation explains the image's transfer, scaling, flipping, tumbling, and shearing. This is how the transfer function appears:

a′ and b′ are indeed the image's true and anticipated classifications, respectively; r denotes the number of categorisations. In this research, the (L-BFGS) art algorithm is utilized to accelerate the classification accuracy as in an optimization function; thus, each parameter update is evaluated using the following.

Translation function:where h  is the current number of training cycles, rounded down. The revolution function is defined as cross entropy, and the formula is as follows.

Revolution formula:

Because of cos  α  and sin  α, there are a limited number of instances in the training set of data, and so this research focuses on the zoom represented in Equation 4 which is treated as training dataset and the p represents the correct ratio which might be too high. In this study, the model's starting correction rate is 30%, and the corrective premium also changes by 80% after two training cycles, as shown in equations (4) and (5):

Zoom:

Scratch:

The compressive transformation matrix is equal to the summation of (5) tan  a  and tan  λ vertical and lateral cracking, as shown below.

The conventional nonlinear part of Equation (7) represented as sa+db+d, ka+f+w  will take on a different affine matrices to achieve the necessary initialization result.

Because acquiring s and  d is a nonlinear process, U is determined to be added to the general iterative procedure. The analysis method is represented by f. For the stopping analysis, k is specified. The continuous analysis was denoted by w.

Each dynamical parameter U also modifies distinguishable aspects of a network, such as its form (in (8)). The transition effect ∑uU(sa+db+d, ka+f+w)  is amplified further by (9) assigning different weight coefficient u to a dynamic emotional attachment property  U:

The aa+λx+ŋ, σa+θb+τ  dynamical transformation is performed as a result of this iterative procedure after conducting a linear transformation on a series of data points (in (9)). As a consequence of both sa+db+d, ka+f+w  wavelet transform and continuous wavelet transform, a comment that can control places is provided. In (10), a function O(a, b) is determined from the dynamical result's first iterative technique.

The O technique iteration would be based only on the major improvement indicated in the preceding section. The procedure formula is as follows:

In a single simple iterative process, its data are first convolved, and then σ is subjected to the wavelet transform. Finally towards the second iterative approach, θ is known as post-transformation. The data analysis is performed by forwarding and commences with reversing the signal propagation. The aggregated nonlinear functional could be configured to manage the transform consequence of such wavelet transform, and the nonlinear transition coefficients u could also be adjusted to match equation.

The data analysis is performed by forwarding and commences with reversing the signal propagation. Because the mass and limit adjustment of each hidden neuron layer surface affects whether the reaction error meets the criteria during back process, scenario is based, i.e., numerous alternative versions, is expected to respect the following equation:

Conventional sin  θ+cos  θ  are termed as digital elements, such as recorders and projectors, are one of the most commonly utilized terms in modern art teaching with architectural painting for education; this d1∗σ3+d2∗σ2+d3∗σ equipment could also be (b, a) employed in online art teaching with taking architectural painting education following the equation below.

Students could study knowledge more (ln(b)) methodically and swiftly only with accompanying computer adjustments (1+b2); quite an educational technique could indeed expand the number of learners (15), educate such much material U2(b, a) in a short period of time, and maximise teacher quality.

The purpose of adding AI to art teaching while adopting architectural painting instruction is to show art content to learners in a much more informed way, to create a suitable educational environment, and also to show preliminary designs that combine technology using the following equation:

3. Results and Discussion

AI performance measurement of modern art teaching is represented as in r00∗a+r01∗b+r02. In the traditional double function iteration, the sequential transformations model is applied. However, in a different perspective on r10∗a+r11∗b+r12  is being used to complete all nesting boxes for self-similarity. Using a 3 × 3 matrix, the results are represented in Figure 2.

Figure 2

Performance Folk music modern art cooccurrence matrix extraction features.

A single iterative basic approach is employed for transformation and nonlinear modifications, and in practice, integration of identity and auxiliary translational and rotational movements are applied. Before choosing the best image, the founder matrix is utilized to prepare that image for numerous modifications. The L-BFGS cost optimization method is then used to increase image quality while reducing noise. The picture extracted characteristics from coaching are compared multiple times to avoid the topic of highly tilted images to incorrect weights and summarize the appropriate style and its loading duration on the Internet. However, altering the variables instantaneously will render multiple kinds of fractal images that attempt to break its single image teaching and try to take architectural painting form and realize the complementary growth of spectroscopic visual art

This pattern is a recurring occurrence in nature and person design, including abstract thoughts. As a result, pattern elements are in a predictable manner. A geometric pattern seems to be a type of pattern made up of geometric shapes that is frequently repeated, such as a design. Patterns can be observed immediately by any sense. The unique pattern qualities of parameters a and b are both greater than 2, meaning that the multi-texture is now simple, consistent, and hence suited for digital picture processing, as shown in Table 1. The photos that are suitable for mixing are sorted out, and the following splicing step is carried out. There are dispersed repetitions and also has continuous cutting edge steps as demonstrated in the conventional format: sa+db+d, ka+f+w  which is a nonlinear format that takes different affine matrices to achieve the necessary values as in Equation 7 with certain initialization result G(a, b)=(sa+db+d, ka+f+w). Because acquiring the values for sand d is a nonlinear process, U is added to the general iterative procedure and is represented in Figure 2. To accomplish the self-affine, unrestricted fineness, and impartial pattern characteristics of spectroscopic graphics, this type of visual can be spliced by utilizing random values for the parameter; the results are obtained as in Table 1, which requires distinguishable pattern characteristics along with regular distribution and can also be dispersed freely. The performance of the pattern characterization can be spliced endlessly.

Table 1

Result analysis of art teaching with architectural painting co-occurrence matrix extraction features.

Parameter	Time (s)	Distinct pattern characteristics	Regular distribution
A	0.825	0.947	5.56
B	0.662	1.403	4.92
R	0.486	1.958	1.978
H	0.525	2.662	2.656

A: distinct pattern characteristics. B: pattern characteristics of spectroscopic graphics. R: pattern characteristics along with regular distribution. H: general unique pattern quality.

AI technology implementation has not yet developed enough. Perception of technology by students, simulation model art teaching and architectural painting instructor innovation, effective educational companion in crime technology, and other technologies that people desire to achieve in the area of art teaching are now being investigated. AI can help individuals to some level but can replace our thoughts, which are the distinctive identifiers of humans. Modern artificial intelligence arts integration cannot reach flawless meanings, attitudes, or representational analyses. Art forms vary and move with the times, but still the significance of art remains constant. Each dynamical parameter U also modifies distinguishable aspects of a network, such as its form (in (8)). The transition effect is amplified further by assigning different weight coefficient u to a dynamic emotional attachment property  U , and the results are updated as in Figure 3. Artworks are usually known as qualitative thinking more in terms of creation and recognition; thus, AI is deficient in the assessment of student art production in terms of visual appreciation. The present AI-based art teaching is just a simple realization of individualized spaces; it simply involves the use of computers to accomplish fundamental art teaching and also can reflect the benefits of AI application in art teaching with architectural painting integration. Only by trying to break the choke point of the mixture of AI with art teaching along with the architectural painting, the people will be able to fully utilize the AI's enormous potential for art education/learning.

Figure 3

Statistics of art modern music using the intelligent online network assisted teaching system.

Even if traditional AI has improved students' academic achievement and passion to some extent, it does not fully comprehend all students' learning and therefore can vary for every individual, nor can it assure that every student engages in instruction. The aa+λx+ŋ, σa+θb+τ  dynamical transformation is performed as a result of this iterative procedure after conducting a linear transformation on a series of data points. As a consequence of both sa+db+d, ka+f+w  wavelet transform and continuous wavelet transform, a comment that can control places is provided. In (10), a function O(a, b) is determined from the dynamical result's first iterative technique. As a result (shown in Figure 4), if people truly want educators to be more actively involved in teaching, and also gather differences in students' educational experiences and provide new reference direction though the private communication, humans must make technological advances in AI which incorporates digital methods and provide greater technical help for education system.

Figure 4

Performance analysis for the statistics for modern art online teaching system.

People acquire data, investigate it, and then extract the information needed to produce Table 2, as shown in Figure 4. Humans might learn from of the data shown in the table in a number of ways by analysing it: Folk music datasets are quite satisfied with option, accounting for almost 97% of the overall population of educators, demonstrating that our inquiry was much more successful and this framework is far more efficient.

Table 2

Result analysis for statistics for modern art online teaching system.

Parameter	Time to paly	Distinct pattern characteristics	Regular distribution	General (%)	Satisfied (%)	Unsatisfied (%)
Art 1	00 : 01 : 18	1.445	1.733	5	89	60
Art 2	00 : 02 : 13	1.843	1.663	3	83	40
Art 3	00 : 01 : 14	0.984	0.757	4	92	66
Art 4	00 : 01 : 13	0.974	0.963	2,7	93	55
Art 5	00 : 00 : 40	0.846	0.558	4,2	85	66
Art 6	00 : 00 : 47	0.423	0.332	3.6	76	53
Art 7	00 : 02 : 36	1.683	1.857	3.3	89	64
Art 8	00 : 00 : 44	0.415	0.474	1.63	71	44
Art 9	00 : 02 : 70	1.854	1.958	4	72	55
Art 10	00 : 02 : 32	1.735	1.553	3	73	4

Standard Internet components, such as recording equipment and video screens, are the most commonly utilised digital means in current education; nevertheless, such equipment can also be employed in home art instruction. Conventional sin  θ+cos  θ  digital elements, such as recorders and also projectors, are one of the most commonly utilised in modern art teaching with taking architectural painting education; this d1∗σ3+d2∗σ2+d3∗σ equipment could also be (b, a) employed in online art teaching with taking architectural painting education based on to retrieve in Figure 5. Teachers might use images, films, and some other media to bring art lessons and architecture painting to life. Using the folk music dataset for teaching as an example, students could acquire knowledge by recognizing the music patterns more systematically and quickly with online education. The learning technique can increase the number of learners, teach more content in a given time, and increase the effectiveness of educators. The goal of integrating AI into arts is to display art teaching and consider architectural painting material to the learners in a much more informed way. This teaching method will build a unique educational environment by illustrating design that incorporates art and science to support the students who enter as creators of talented artists. Truly engage learners within masterworks of talented artists to break the constraints of place and time and assist viewers in noticing. However, a shortage of AI device capabilities in art education will not achieve the desire for art teaching. Architectural painting teaching results in good outcomes or goals that are difficult to achieve.

Figure 5

Accuracy analysis table of convolutional L-BFGS art algorithm.

The classification performance of the interpolated L-BFGS method accuracy frame is 89.54%, and the prediction accuracy is 87.23 percent, according the data shown in Table 3. The first type of fault (unexpected data is viewed as mistakes) of the training and test set is 75.93% and 69.86%, respectively, and the second type of fault (regular data is viewed as different data) is 83.76 percent and 79.95%, including both, indicating that woman's authenticity could be guaranteed to a certain extent.

Table 3

Result of accuracy analysis of convolutional L-BFGS art algorithm.

Sample	Experiential learning	0	Forecast	Correct rate (%)
Performance of training	0	98	21	85.57
	1	22	84	72.93
	Total proportion	74.23%	64.34%	84.78
Performance of testing	0	7	3	88.22
	1	3	6	67.88
	Total proportion	71%	27%	78.96

Nonetheless, many existing researches have concentrated on the deployments of implementation in art teaching, with architecture design in the art education, and on standard impact difficulties in such a specific link of art education or creation procedure. It also ignores the overall planning process of AI technologies through art teaching. Students could study knowledge more (ln(b)) methodically and swiftly only with accompanying computer adjustments (1+b2); quite an educational technique could indeed expand the number of learners, educate such much material U2(b, a) in a short period of time, and the result of maximizing teacher quality is mentioned in Figure 6. To that end, this research analyses the use of artificial intelligence for artistic teaching including architectural painting education from the viewpoints of strategic planning and model development utilising inductive approach. This study has evaluated the existing algorithms and compared them with the L-BFGS algorithm correctness framework, with the latter obtaining the greatest performance results of training (97%) and testing (98%) (refer to Table 4).

Figure 6

Performance and comparison of various algorithms.

Table 4

Performance result analysis for comparison of various algorithms.

Algorithm	Training (%)	Testing (%)	Overall accuracy (%)
L-BFGS	97	98	98.76
Gradient Descent	86	84	87.45
Adam	78	69	76.23
Adadelta	79	81	80.12

4. Conclusions

Because of the rapid growth of technology today, artificial intelligence has appeared in virtually every field. Reduced human intervention and improved results are two of the artificial intelligence's primary goals. In this project, researchers examine how artificial intelligence might enhance art instruction, using architectural painting as a case study. Artwork that depicts only architecture, including indoor and outdoor vistas, is architectural painting. Paintings in the early stages tended to display building only as a background feature rather than as the primary focus of the work. When painting began to focus more on architecture, it became popular. Internet technology, wireless sensor networks (WSNs), and artificial intelligence technologies like deep learning, as demonstrated previously, are all being used in art education.

As a result of AI, education has become much more manageable. Internet, wireless sensor networks, artificial intelligence, and lightweight deep learning models are all part of the system that we are proposing for art education. Adopting this new technology improves the teaching technique. People use the Limited Broyden–Fletcher–Goldfarb–Shanno (L-BFGS) art algorithm for the suggested system analysis compared to established algorithms like Gradient Descent, Adam, and Adadelta. The proposed algorithm attained 97 percent and 98 percent accuracy in both the training and testing phases, as can be seen from the data.

However, gaining architectural painting knowledge through intelligent art classes will not be sufficient in the current scenario. Hence, the candidates have to gain exposure through real-time painting designs by various painters and their strategies. In this context, as a future enhancement, an intelligent system can be designed to collect these real-time paintings by the painters and give a broadened knowledge to the candidates.