Introduction

Efflorescence is one of the most common diseases in brick cultural relics. The primary cause of efflorescence is the exposure of brick masonry to water, wind, and freeze-thaw cycles, all of which are closely tied to climatic conditions. Different temperatures, humidity, different wind and hydrological characteristics will cause certain erosion to brick cultural relics, especially causing different degrees of efflorescence to cultural relics. Wakana Araoka et al.1 believed that factors such as microorganisms, salt crystallization and masonry cracks would cause the Nanjing city wall to deteriorate, and proposed a simulation model to represent rainwater infiltration into the city wall. Based on the simulation results, they examined the impact of rainwater on the deterioration of the city wall. Yan Ma et al.2 believed that a suitable humid-thermal environment is essential for the sustainable protection of movable and immovable cultural heritage. They proposed a method that emphasizes the combination of humid-thermal coupled transfer simulation and traditional environmental monitoring for systematic risk assessment to quantitatively assess the water-related deterioration risks in heritage buildings. This method has been verified in the Forbidden City of China, a UNESCO World Heritage Site. This study aims to detect efflorescent areas in brick cultural relics using image recognition technology and machine learning, and to explore the influence of meteorological parameters on the walls of brick-concrete structure buildings in hot summer and cold winter areas from a macro perspective.

Central China has a northern subtropical monsoon climate characterized by abundant rainfall, sufficient heat, synchronized rain and heat, as well as cold winters and hot summers. Therefore, all kinds of buildings built in this area face serious moisture damage3,4,5. In Wuhan, bricks have a long history of being used as building materials. Initially, people repeatedly beat clay into a blank and consciously mix straw to enhance the strength of the blank, thus forming the earliest adobe bricks. Until modern times, when Western powers opened China’s door, the red brick materials advocated by Western architectural construction also took root in Wuhan, thus opening a prelude to the use of red brick materials in modern Wuhan6,7 Therefore, bricks were chosen as the research object for the efflorescence phenomenon.

The impact of moisture is more evident for brick buildings built in a humid environment, the impact of moisture is more obvious. A large amount of moisture entering the interior of the brick wall will causes the brick wall surface to become efflorescent8,9,10. In the long term, efflorescence on the brick wall surface will directly reduce the bearing capacity of the brick wall structure. After the brick wall surface peels off owing to efflorescence, more air, moisture and salt in the environment will enter again, causing corrosion of the brick wall, and ultimately leading to structural failure and even major accidents11,12,13 To prevent structural failure due to moisture, timely and effective reinforcement and repair of brick wall damage are essential, and the primary task of brick wall reinforcement is to identify the efflorescence area on the brick wall caused by moisture diseases14.In the past, the detection of building surface defects relied mainly on manual inspection. However, manual inspection is time-consuming, subjective, and sometimes dangerous. Due to the efficiency and accuracy of deep learning, computer-vision-based defect detection is gradually replacing manual visual inspection15,16. Therefore, computer vision-based identification of damp defects in brick walls is crucial. At present, the research object of brick damage recognition based on computer vision is mainly based on brick wall cracks. Research on intelligent crack detection primarily includes three aspects: crack classification, bounding box regression, and semantic segmentation. Cracks can be classified into several types, and each type is categorized into different categories17,18,19,20 Bounding box regression is used to achieve the target detection of cracks, and the target detection result will show the bounding box of crack ___location and the confidence level of crack identification21,22. Dongho Kang et al.23, proposed an autonomous UAV method using ultrasonic beacons to replace the role of the original GPS, this study mainly used a geo-tagging method for damage localization and deep convolutional neural networks to identify the damage of various types of building structures, which breaks the tradition of the original use of visual inspection for structural health monitoring. Rahmat Ali24 et al. proposed an autonomous unmanned aerial vehicle (UAV) system integrated with a modified Faster R-CNN to recognize various types of structural damage and map the detected damage to GPS-less environments.

Existing literature on the identification of damage to brick walls has mainly focussed on cracks, with little research on damp damage to brick walls. However, damage to brick walls from efflorescence is inevitable, particularly because a wide range of buildings constructed in wet environments is constantly affected by efflorescence. In this study, brick wall flooding was investigated and a brick-wall flooding dataset was generated. Deep learning was used to identify brick-wall-efflorescent damage. The main contributions of this study are as follows.

①This study conducts a systematic study on the intelligent identification of alkali efflorescence areas in brick walls and creates a dataset of brick-concrete building wall efflorescence in Wuhan City, Hubei Province, China. The target detection model trained on the dataset can be applied in practical applications.

② Through the latest target detection method (Yolov10), the area of efflorescence in the brick wall is determined, and then the cause of efflorescence is inferred based on the climate characteristics of the area where the building is located.

③ It is suggested that real-time monitoring should be implemented for cultural relics buildings with brick-concrete structures, so as to effectively curb the further deterioration of efflorescent disease in brick cultural relics buildings. At the same time, two constructive suggestions were put forward for the removal of efflorescent areas on brick walls.

Methods

Study area

Wuhan City is located in the eastern part of Hubei Province (Fig. 1). According to the provisions of the “Building Climate Zoning Standard” GB50178-93, Wuhan is located in a region with hot summers and cold winters and has a typical subtropical humid climate.

Fig. 1: Study area.
figure 1

The main study area of this research is located in Jiangan District, Wuhan City, Hubei Province, China(Map source: This map is produced based on the standard map with approval number GS (2019) 1673 downloaded from the National General Geospatial Information Service Platform Standard Map Service Website. The base map has not been modified.).

Full size image

After the signing of the Treaty of Tianjin in 1858, foreign missionaries were allowed to enter China to carry out missionary activities, and Western-style hospitals took root in Wuhan today. Dr. Porter Smith, a British physician, was dispatched by the Methodist Church of England to establish the first Western-style hospital in Wuhan, the Hankou Pouai Hospital. Thereafter, the London Society of Christ in England, the Baptist Church in England, the Episcopal Church of Christ in America, the Church of Christ in China, and the Italian Canonical Sisters of Charity of the Catholic Church and the North American Sisters of Charity of the Catholic Church successively hosted the establishment of church hospitals in Wuhan for the purpose of missionary work. Since the Revolution of 1911 and the Republic of China, the government has begun to attach importance to public health and Western medicine. The government and various organizations have presided over the establishment of public hospitals. During this period, many doctors have presided over the establishment of private hospitals. In addition, many military hospitals appeared in Wuhan due to the outbreak of the Revolution of 1911, the Northern Expedition, the War of Resistance against Japan, and the War of Liberation. This article collects the most valuable 98 modern medical buildings in Wuhan from 1861 to 1949 in Wuchang, Hankou, and Hanyang, and conducts an in-depth study on them25,26,27,28,29 The study found that modern Wuhan hospitals can be firstly divided into three categories: foreign-funded hospitals, domestic-funded hospitals and military hospitals. Among them, foreign-funded hospitals can be further divided into three categories: Christian church hospitals, Catholic church hospitals and other hospitals. Domestic-funded hospitals can be divided into two categories: public hospitals and private hospitals, and the statistics of the number of hospitals of specific types of nature are as follows (Table 1). Until 2024, with the changes in the medical industry and the development of Wuhan, the layout of modern medical buildings has long been unable to meet the needs of current patients, the medical buildings listed in the table below have almost been demolished, with only 12 remaining, but only as a commercial building or a historical building. Although Gao’s Hospital is not a large building, it is the only existing private hospital in Wuhan founded by a female doctor, which has unique cultural relics, historical and artistic values in the history of modern architecture in Wuhan. Therefore, this paper takes Gao’s Hospital as the main research object to identify and detect efflorescence on the exterior walls of its building.

Table 1 Statistical Table of Wuhan Modern Hospital (1861–1949)
Full size table

Gao’s Hospital was founded by female doctor Gao Xinrong. Gao Xinrong was born in Huarong Town, a generation along the Yangtze River, in Echeng County, Hubei Province (now Ezhou City) in the late Qing Dynasty. After the Revolution of 1911, Gao Xinrong’s father followed Sun Yat-sen to Wuhan, and Gao Xinrong entered a private school. At the age of 18, Gao Xinrong was admitted to the Summergate Medical School founded by the American Christian Church. In the summer of 1936, Gao Xinrong went to the United States alone and entered the prestigious Johns Hopkins University School of Medicine in Maryland on the east coast of the United States to begin a one-year training. After a year, after completing the theoretical training, she decided to study surgery. Recommended by professors at the school, she entered the pathology teaching and research section of the American-Russian Research Institute of the University of Minnesota. The institute gathered famous experts and scholars from the United States and Europe, and collected the world’s difficult and rare cases and complex surgical and gynecological operations. One year after she went abroad, the Anti-Japanese War broke out. One year after going abroad, the Anti-Japanese War broke out. According to the plan, she planned to study at the American-Russian Institute until the end of the year, and then go to Zurich Medical College in Switzerland for another year. One day in April 1939, a British pastor came to the American-Russian Institute to give a speech. He told about his recent experience in China, exposing the atrocities of the Japanese army and the tragic conditions of the war. The pastor shouted: “Everyone come to rescue China! The Chinese people are fighting against the Japanese fascists with their flesh and blood. Every moment, loyal and brave sons are bleeding and sacrificing. Medical personnel are urgently needed there!” Gao Xinrong was shocked when she heard this. A foreigner was still running around for China’s war of resistance. As a Chinese, how could she stand idly by? “Every man has his share of responsibility for the rise and fall of the country.” She seemed to hear her father’s voice. She immediately decided to return home.When Gao Xinrong arrived in Hong Kong, China, Liu Ruiheng, director of the National Government’s Department of Health, met with her and hired her to work at Chongqing Central Hospital. Due to frequent large-scale bombings by Japanese warplanes, most of Chongqing’s offices, factories, and citizens had moved to the countryside. At this time, Chongqing Central Hospital was being built in Gele Mountain, so Gao Xinrong was assigned by the Department of Health to work at the Sansheng Palace Health Center, which was more than 10 kilometers away from the city. There was only one doctor in the health center, Gao Xinrong, and she was the only one who saw patients in internal medicine, surgery, gynecology, and pediatrics. It was common for her to be busy from morning to night. Guo Moruo and Yu Liqun and his wife lived very close to each other, and Yu Liqun often came here to walk around or see a doctor. Two years later, the Central Hospital was completed, and Gao Xinrong was ordered to be transferred to the deputy director of the hospital’s obstetrics and gynecology department. During this period, Gao Xinrong won the respect of her colleagues with her solid foundation and excellent skills. On August 15, 1945, Japan surrendered unconditionally. In the winter of 1945, Gao Xinrong returned to Wuhan by boat. She and her younger brother Gao Youbing were employed as the director of the Department of Obstetrics and Gynecology and the chief physician of the Department of Surgery at Hankou Municipal Hospital respectively. Soon, her older brother Gao Youhuan returned to Hankou from Yunnan and also worked at the Municipal Hospital.After arriving at Hankou Municipal Hospital, Gao Xinrong urged the hospital to purchase necessary equipment and establish relevant medical systems, so that the hospital’s work could quickly get on track. In order to improve the technical level of medical staff, she started from the basics, strictly required medical, nursing and midwifery staff, and patiently explained and taught advanced medical technology to doctors and nurses in combination with cases during daily rounds. After the end of the Anti-Japanese War, the degeneration of the atmosphere of Hankou Municipal Hospital became a more serious problem in its development process. At that time, many members of the municipal government criticized Hankou Municipal Hospital for being bureaucratic, doctors being bureaucratic, and treating diagnosis as a perfunctory official business, which repeatedly delayed the lives of patients. After 1948, the National Government had a serious fiscal deficit and excessive currency issuance, which led to inflation and soaring prices. In order to make up for the lack of living expenses, the employees of Hankou Municipal Hospital took various ways to make extra money. For example, the drugs prescribed by doctors for patients had to be purchased by patients at designated pharmacies. Gao Xinrong was dissatisfied with the practice of some people in the hospital who were greedy for windfalls, so she resigned and went home with her two younger brothers. Gao’s Hospital was officially opened29 (Fig. 2).

Fig. 2: Gao Xingrong's Biography.
figure 2

a Dr. Gao Xinrong performing gynecological surgery; b Doctor Gao Xinrong’s application to join the Communist Party of China.

Full size image

Fig. 3 illustrates Gao’s Hospital is located at No. 38, Lihuangpi Road, Jiang’an District, Wuhan City, Hubei Province. It was established in 1936. The building is a three-story brick-concrete structure in the Baroque architectural style. Red fired bricks are the main building materials, granite and cement are used as the foundation, and the arched doors and square windows are mainly made of cement25. The hospital is designed as a rectangular plan, with three departments of internal medicine, external medicine and gynecology and obstetrics, with 18 in-patient beds, mainly for maternity hospitalization. The ground floor of the hospital for the outpatient department, pharmacy, laboratory, injection room and small operating theatre, small operating theatre can be appendix, curettage and other operations, in the event of ectopic pregnancy, ovarian cysts, uterine removal and other major operations, they will be across the street with the help of the operating theatre of the World Wide Web (now Wuhan Hospital of Traditional Chinese Medicine). The second and third floors of the building are the inpatient building on one side and the Gao family residence on the other side. The facade is free and lively. The facade along the street adopts an asymmetrical layout with concave and convex changes. The entrance side is curved. There are decorative pilasters between the first and second floors, and the third floor is a white double pilaster arch window. The exterior walls of the building are decorated with bricks, and the roof is a sloping roof made of tiles. Gao’s Hospital has important historical value in terms of culture and artistry, and has important reference value for the study of the architectural style of the concession area during this period. Gao’s Hospital officially closed in 1952. Gao Xinrong and Gao Youbing went to Wuhan Second People’s Hospital to preside over the obstetrics and gynecology and surgery departments, and donated all the equipment and medicines of Gao’s Hospital to the country free of charge. Gao’s Hospital is now used as a store and art gallery (Meilian Society). The building has been well protected and was announced as an excellent historical building in Wuhan in 2007 (Fig. 3).

Fig. 3: Location of Gao Hospital.
figure 3

a The ___location of Gao’s Hospital in the Russian Concession in Hankou; b Aerial view of Gao Hospital.

Full size image

Fundamentals

Yolo (You Only Look Once) is a single-stage detector that performs localization and classification tasks in a single pass of the network architecture. The main architecture of Yolo is divided into three main components: (1) The backbone contains a series of layers (convolution, pooling) and blocks (C3, SPPF, C2F). The backbone spatially down samples the input image while adding feature maps. It captures hierarchical image features at different scales and resolutions and acts as a feature extractor, with the initial layer extracting low-level features and the deeper layer extracting high-level features; (2) The neck forms a connection with the head to form a series of intermediate layers including convolution, spatial pyramid pooling (SPP) and feature pyramid network (PANet). It refines the features extracted from the backbone and enhances the spatial and semantic feature representation at different scales to handle objects of different sizes; (3) The header contains multiple convolutional layers and activation functions that make final predictions (i.e., bounding box and class probabilities) based on the features extracted from the backbone and neck. In the post-processing stage, it performs NMS operations to refine the final detection.

Introduced in 2015, Yolov1 was seen as a new approach to target detection by achieving good accuracy and speed by processing images in a single stage. The first Yolo version laid the foundation for real-time applications and set a new standard for subsequent developments30.Yolov2 (Yolo9000), an extension of Yolov1, increased the operational resolution of the system to be able to detect more than 9000 object classes, thus enhancing its versatility and accuracy31. Yolov3 further enhances versatility and accuracy by implementing multi-scale prediction and a deeper network architecture for more accurate detection of small objects32. The series continued to evolve with the introduction of Yolov4 and Yolov5, each of which introduced finer-grained techniques and optimizations to further improve detection performance (i.e., accuracy and speed). Yolov4 incorporated features such as cross-stage-part (CSP) connectivity and mosaic data enhancement, while Yolov5, developed by Ultralytics, offers significant improvements in terms of ease-of-use and performance. significant improvements, making Yolo a popular choice among the computer vision community33,34. Subsequent versions, Yolov6 through Yolov9, continued to build on this foundation, focusing on increasing model scalability, reducing computational requirements, and improving real-time performance metrics35,36,37,38 (Fig. 4).

Fig. 4: YOLO development history.
figure 4

The target detection model has been iterated for ten generations from 2015 to the present (2024).

Full size image

Figure 5 shows Yolov10 is the latest version of the Yolo series. It was developed by researchers at Tsinghua University based on the Ultralytics Python package and will be officially launched and put into use in May 202439 Yolov10 perfectly solves the post-processing and model problems of previous versions. Due to architectural deficiencies, Yolov10 optimizes the model architecture and eliminates the NMS maximum suppression operation, which not only significantly reduces the overhead required for model operation, but also raises the performance of its target detection to the highest point.

Fig. 5: Yolov10 framework.
figure 5

The latest generation has introduced a dual label assignment mechanism.

Full size image

Wall efflorescent data set

The dataset used for training contains 1986 pictures of brick wall efflorescence. Some of these pictures come from the Internet, and the other part is taken by the author with a smartphone and camera. The pictures were annotated and classified using the Roboflow platform. 66% of the pictures are allocated for training, 27% are reserved for verification, and 7% are used for evaluation. In order to ensure the generalization and robustness of the target detection model, the author has performed data augmentation on the training set in the dataset. The color, brightness, and shooting angle of the brick wall efflorescence pictures in the training set are different. The original database samples used for training, testing, and evaluation are shown in Fig. 6. The detailed information of the dataset is shown in Table 2.

Fig. 6: Database example.
figure 6

There are 1986 images in the data set, which contain different styles and varying degrees of efflorescence.

Full size image
Table 2 Basic information table of Brick wall efflorescence dataset
Full size table

Learning frameworks

This section details how the object detection model Yolo operates. Figure 7 describes the training and testing workflow in deep learning. All training was performed on a desktop computer with a single GPU NVIDIA GeForce RTX 4060 with 8 GB of memory (Table 3). First, the image and annotation data were merged into the same file and input into the Yolo network architecture for training. The initial settings before training were as follows: initial learning rate 0.001; mini-batch size 20; threshold 0.5; and epoch 300. During training, we found that the training loss initially remained at a high level based on the loss curve, but then the training loss also decreased with the increase in the number of training times, and the decrease was significant. During the test process, the specific ___location of the wall efflorescence was represented by the bounding box of the detector. Finally, with the end of training and testing, the accuracy and speed of the YoloV10 algorithm in wall efflorescence recognition were shown.

Fig. 7
figure 7

Workflow of YoloV10 for Wall Flooding Training and Testing.

Full size image
Table 3 Model training environment (Desktop Computer Hardware and Software Configuration)
Full size table

Results

Results of pantethine testing conducted at Gao’s Hospital

Due to laboratory environment and time constraints, we finally decided to use the YoloV10n model in the YoloV10 series to conduct this experiment. Table 4 presents in order to verify the effectiveness of the Yolov10n model, we commonly use four indicator tables (precision, recall, F1 score and mAP) supplemented by a confusion matrix to measure the performance of the model. At the same time, we also set a confidence threshold (0.25), based on the experimental results impartiality and objectivity40.

Table 4 300 rounds of model training performance parameters
Full size table

The accuracy of the model demonstrates a rapid improvement during the initial training phase. After 100 trainings, the accuracy began to fluctuate periodically, and gradually stabilized in the final stage of learning. This indicates that the model continuously processes different data during the learning process and determines the local optimal solution at a certain node. After the training was completed, the accuracy of the model was mainly stable in a relatively high range (0.857–0.907), which shows that Yolov10n has absorbed the main information of wall efflorescence (Fig. 8a).

Fig. 8: Model Operation Results.
figure 8

a Performance Parameters of Model Training; b Precision-Recall Curve.

Full size image

The recall rate of the model experienced a fluctuation during the training process. At the beginning of the training, the initial value of the recall rate increased from 0.48345 to 0.56957 and then gradually decreased to 0.49497. As the model accuracy began to improve gradually, the recall rate also showed an upward trend until it reached 0.86522 (Fig. 8a).

During the training process, the three loss functions continued to decline, which means that the performance of the YOLOv10n model is extremely excellent in all aspects. box_loss, which represents the accuracy of the bounding box, dropped from 3.4622 to about 2.5017, which means that the model’s ability to locate objects in images is significantly improved; similarly, cls_loss, which reflects the accuracy of object classification, also dropped significantly from 7.0811 to 1.4081, which shows that the model is improving significantly. Proficiency in correctly identifying objects continues to improve; dfl_loss represents the distribution fit of the model, and its value has dropped from 4.5085 to 3.4184, indicating the model’s increased ability to handle complex data distributions or refine predictions (Fig. 8a).

Figure 9b presents the F1 vs. confidence curve shows the F1 score of the model at different confidence thresholds. The experimental results show that the modified curve of Yolov10n has a higher f1 value at the maximum confidence threshold, indicating that the model performs very well at different confidence levels. The curve of the model was relatively smooth, indicating that the performance of Yolov10n at different confidence levels was more stable. When the confidence threshold was 0.31, the f1 value reaches 0.83. The value range of [email protected] increases from 0.4 to 0.898.

Fig. 9: Results after Target Recognition.
figure 9

a Identification Results of Wall Flooding in Gao’s Hospital; b Wall Flooding Areas in Gao’s Hospital.

Full size image

Efflorescence of walls at Gao’s Hospital

Efflorescence typically occurs alongside the chalking of masonry blocks owing to water erosion. Gao’s hospital is located in an area with hot summer and cold winter although the soil humidity and air relative humidity are large, because the building adopts a stone foundation, which plays a certain role in isolating underground and surface water from moisture, the efflorescent areas of the building are primarily concentrated on the exterior walls beneath the windowsills, eaves, and around drainage pipes, the area where the drainage pipe flows through, the faucet So the alkali-flooded areas of the building are mainly concentrated under the window sills and eaves of the building’s exterior walls, in the areas where the drainage pipes flow through, in the areas where the water faucets are installed, and in other wall surfaces that may be attacked by rainwater, with a larger area of alkali-flooded surfaces on the backs of the sunny walls (Fig. 9). The author also determined the extent of efflorescence of the walls of Gao’s Hospital through the frost test in Chapter 11 of the National Standard of the People’s Republic of China “GB/T 2542-2012 Test methods for wall bricks” (Table 5)41. Although the area where alkali seepage occurred on the wall of Gao’s Hospital is large, it is located in the historical city area of the former concession of Hankou, which is a first-level style control area designated by the Wuhan Municipal Government in the Wuhan Historical and Cultural City Protection Plan (2022–2035). In recent years, it has been referring to the relevant laws and regulations on the control requirements of the core protection areas of historical and cultural style blocks and the ecological bottom line areas, so the overall part with the most serious alkali seepage has only reached a moderate level.

Table 5 Description of the degree of wall efflorescence
Full size table

Analysis of influencing factors of wall efflorescence

After on-site investigation and target detection, it was confirmed that the reasons for the occurrence of efflorescence in Gao’s Hospital can be divided into two aspects, namely the original material of the wall and the local meteorological parameters of Wuhan.

Raw material impact

Table 6 juxtapose the phenomenon of efflorescence often occurs because the soluble substances in the cement, sand, gravel, bricks, blocks and admixtures used as raw materials are dissolved by water. After dissolution, they will gradually precipitate out of the building as the water in the blocks and mortar evaporates. On the surface, it reacts with carbon dioxide in the air to form a white deposit42.

Table 6 Chemical composition of wall flooding
Full size table

Impact of meteorological parameters

The influence of meteorological parameters is caused by local hydraulic erosion, freeze-thaw erosion, and wind erosion in Wuhan, that is, water, temperature, and wind are the main factors causing alkali efflorescence of the walls of Gao’s Hospital. The moisture inside the wall as the building envelope mainly comes from the water vapor exchange between the wall and the outside, atmospheric precipitation, and soil moisture. The water vapor exchange between the wall and the outside and atmospheric precipitation can be reflected by the relative humidity data of the air: the moisture of the soil mainly comes from surface seepage and groundwater. In areas with high groundwater levels, the waterproof and pull-out resistance of the building foundation are required to be high. Modern brick cultural relics buildings usually do not have moisture-proof layers, and the walls are easily eroded by moisture and become damp. Due to the high water absorption rate of bricks, under the influence of humid environments, the large fluctuations in temperature in spring and winter will cause the moisture inside the brick masonry to freeze and thaw, thereby causing the bricks to be eroded by freeze-thaw; the action of wind can weather the surface of the material. In a humid environment, the wind acts on the bricks after absorbing water, which will aggravate the brittleness and looseness of the material’s microstructure, so that the bricks show efflorescence from a macroscopic perspective. It can be seen that hydraulic erosion is the prerequisite for freeze-thaw erosion and an accelerator of wind erosion. Therefore, the main influencing factors of building alkali-fat disease can be attributed to three aspects: hydraulic action, freeze-thaw action, and wind action43,44.

  1. 1.

    Hydraulic erosion under the action of relative humidity and water table

    Gao’s Hospital is located in Jiangan District of Wuhan City, its core area is at latitude 29°58′N, longitude 113°41′E’, belonging to the northern subtropical monsoon humid climate zone, with the characteristics of “more rainfall, less sunshine, weak wind”, with an average annual precipitation of about 1316.2 (mm). Meanwhile, as Wuhan is located in the south bank of the middle reaches of the Yangtze River, there are many tributaries and streams in the surrounding area, and the total area of the river basin can be up to 159,000 square meters, which is rich in water. As a result, the relative humidity of the air is extremely high, usually reaching over 80% (Fig. 10).

    Fig. 10
    figure 10

    Annual Humidity in Wuhan (2023).

    Full size image

    Although the underground water level of Gao’s Hospital is relatively low, the surrounding water system is rich, the precipitation is large, the soil moisture is high, and the relative humidity of the air is high, which makes the wall efflorescent area larger. Secondly, because the main materials of the building foundation are rubble and concrete, they play a moisture-proof role to a certain extent, so the water line of all walls is not very obvious.

  2. 2.

    Freezing and thawing erosion under the effect of temperature

    Wuhan has a mild and humid climate throughout the year, with a rapid warm-up in spring, a long and hot summer with heavy rainfall, a short autumn with frequent precipitation, and a relatively warm and short-lived winter, with the average temperature of the coldest month being about 3 °C, the average relative humidity reaching 77–80%, and the frost-free period being relatively long, at about 240 days. The average annual temperature is between 15.8 °C and 17.5 °C, and the annual temperature is basically above 0 °C, with the extreme lowest temperature occurring in 1977, reaching −18.1 °C. Overall, freezing and thawing rarely occur in Wuhan, and higher relative humidity and wind erosion are the main factors that cause efflorescent diseases in buildings (Fig. 11).

    Fig. 11
    figure 11

    Wuhan Yearly Dry Bulb and Wet Bulb Temperatures (2023).

    Full size image

    It can be seen from this that since the temperature in Wuhan is basically above zero throughout the year, although there are some days of ice and freezing, it is not enough to form a freeze-thaw cycle, so its freeze-thaw erosion can be basically ignored.

  3. 3.

    Sustained low-level wind erosion

Figure 12 illustrates the maximum wind force in Wuhan usually does not exceed 8, and the main wind force levels with the corresponding number of days are 180 days of breeze, 150 days of wind less than level 3, 120 days of wind level 3–4, 60 days of wind level 4–5, 30 days of wind level 5–6, and the rest of the wind force levels are not more than 5 days.

Fig. 12
figure 12

One-Year Windy Rose Map of Wuhan (2023).

Full size image

It can be seen that although the Wuhan area is located in a relatively low-lying area with relatively weak wind, Gao’s Hospital is located along the Yangtze River, so it is still subject to wind erosion from humid air all year round. At the same time, the continuous low wind level also has the effect of water dripping through the stone. In general, this situation has a stronger effect on the weathering of the blocks. When dealing with wall efflorescence in the later stage, you need to consider not only the waterproofing of the exterior wall, but also its wind proofness. For example, proper spraying (water-based silicone emulsion) on the surface of the brick wall with a hydrophobic agent can be used in the waterproofing process. while effectively preventing wind.

Brick flooding repair strategies

China’s modern brick-concrete structure architectural heritage integrates many valuable elements such as history, art, science, and social value, and is an important part of national historical relics. However, in today’s era, many modern brick-concrete structure buildings have efflorescence on the exterior walls to varying degrees due to the influence of the two factors mentioned above and human damage. These efflorescence will have many adverse effects on the wall, so we have proposed two methods based on Chapter 11 of “Test Methods for Wall Bricks GB/T 2542-2012” to remove the efflorescence of the wall:

  1. 1.

    Sandblasting method: Sandblasting: Sandblasting is to use a sandblasting machine to spray dry fine sand on the wall with efflorescence. Under the high-intensity impact of the sand particles, the efflorescence material gradually falls off the surface of the brick wall. This method is simple and easy to implement, but the disadvantage is that it cannot fundamentally suppress the efflorescence phenomenon.

  2. 2.

    Acid cleaning method: Clean with 1:10 dilute hydrochloric acid. First wet the surface to be cleaned with clean water, then clean with dilute hydrochloric acid, and finally clean with plenty of clean water. This method can convert calcium carbonate crystals that are insoluble in water into calcium chloride that is easily soluble in water through chemical reactions. The removal effect is significant, but acidic substances penetrate into the permeable bricks and have a certain negative impact on their mechanical properties.

Discussion

This study explored and implemented a brick wall efflorescence detection technology based on target recognition, applying the most efficient Yolov10 algorithm. Through in-depth analysis of the special structure and working mechanism of Yolov10’s deep convolutional neural network combined with multi-scale detection technology, the advantages and practicality of this algorithm in the field of wall efflorescence detection were demonstrated.

During the research, considering the actual needs of brick wall buildings, a brick wall surface damage alkali efflorescence dataset was specially designed and supplemented to make the model training closer to the actual application scenario. The experimental results show that the brick wall efflorescence area detection model based on Yolov10 proposed in this study has a high accuracy rate of 85.7–90.7% on different types of brick wall surfaces, validating the effectiveness and reliability of this technology in detecting efflorescence in brick walls.

At the same time, the research continued to determine the reasons for the alkali efflux of Gao’s Hospital brick walls due to different meteorological parameters through test results, namely water, wind and freeze-thaw erosion. Among them, wind and water erosion are the most obvious, followed by freeze-thaw erosion.

This study is only to explore one of the damp diseases of brick wall buildings. It is the beginning of the study of building damp diseases. The mechanisms underlying the formation of damp-related building diseases are complex and warrant further scientific investigation: we can expect that the research on building damp prevention methods in the future can lead to an effective way.