AI-Based Leakage Detection System Using Image Analysis and Method Thereof

Abstract

A leak detection system using artificial intelligence-based image analysis and a method of operation thereof are disclosed. A leak detection system according to one embodiment of the present invention may include at least one camera for capturing image data and an image analysis device that analyzes the image data to extract feature information for determining whether a leak has occurred, and determines whether a leak has occurred by comparing the feature information with a pre-trained model.

Description

Technology Field

[0001] The present invention relates to a leak detection system using artificial intelligence-based image analysis and a method of operation thereof. More specifically, the invention relates to a leak detection system and a method of operation thereof that extracts feature information for determining whether a leak has occurred from image data capturing an area where a chemical is leaking in a semiconductor facility, etc., and determines whether a leak has occurred by comparing the feature information with a pre-trained model. Background Technology

[0003] In semiconductor manufacturing processes, various chemicals are transported and processed through pipes, valves, flanges, and the inside of process equipment; if these chemicals leak, they can cause equipment damage, production stoppages, and worker safety issues. In particular, in environments where precision processes are performed, such as semiconductor facilities, even minute leaks can lead to equipment malfunctions or process defects, so technology capable of rapidly and accurately detecting leaks is required.

[0004] Conventionally, the primary method used to detect leaks involves installing sensors at specific locations. Such sensor-based systems require different sensors depending on the type of chemical and necessitate the individual placement of sensors at every location where a leak is anticipated. Consequently, monitoring a wide area requires numerous sensors and wiring, leading to complex installation and maintenance processes. Furthermore, since leaks can only be detected within the specific sections where sensors are installed, there is a limitation in that it is difficult to detect leaks occurring in areas where sensors are not present.

[0005] Furthermore, sensor-based systems face the problem of increased implementation costs due to complex wiring structures and the frequent requirement of additional devices such as controllers and repeaters; there is also a possibility of sensor malfunctions or detection errors caused by environmental conditions. Additionally, real-time response is difficult because administrators often have to physically inspect the site even after a leak is detected.

[0006] Meanwhile, although video-based surveillance technology is being applied to various industrial fields recently, simple video surveillance methods have problems such as difficulty in automatically determining the presence of leakage and the potential for false detections caused by external environmental factors like lighting changes, shadows, and reflections. Therefore, there is a need for a technology capable of extracting leakage-related features from video data and determining the presence of leakage by comparing them with a trained model.

[0007] Accordingly, there is a growing need for leak detection technology that can automatically determine whether a leak has occurred based on video data and distinguish between leaks and non-leaks even under various environmental change conditions. Prior art literature

[0009] Korean Registered Patent No. 10-2516839 (Registered on March 28, 2023) The problem to be solved

[0010] The present invention aims to solve the aforementioned problems by providing a technology that resolves the issues of conventional sensor-based leak detection technology, where the monitoring range is limited because leaks can only be detected through sensors installed at specific locations, and where the installation and maintenance processes become complex due to the requirement of multiple sensors and wiring to monitor a wide area.

[0011] Furthermore, the present invention aims to resolve the problem that conventional leak detection technology makes it difficult to detect leaks occurring in areas where sensors are not installed, and that consistent detection in various environments is difficult due to the structure requiring the use of different sensors depending on the type of chemical.

[0012] Furthermore, the present invention aims to solve the problem in simple image-based surveillance technology where it is difficult to accurately determine whether there is a leak, as false detections may occur due to environmental changes such as lighting changes, shadows, and reflected light when determining the presence of a leak based solely on changes within the image.

[0013] Furthermore, the present invention aims to provide a technology capable of reliably determining leakage even under various environmental conditions by extracting leakage-related features during the process of determining leakage based on image data and comparing them with a pre-trained model to determine leakage.

[0014] The problem to be solved by this specification is not limited to what is described above and can be extended to various matters that can be derived from the embodiments of the invention described below. means of solving the problem

[0016] A leak detection system using artificial intelligence-based image analysis according to one embodiment of the present invention may include at least one camera (110) for capturing image data; and an image analysis device (120) for analyzing the image data to extract feature information for determining whether there is a leak, and for determining whether there is a leak by comparing the feature information with a pre-trained model.

[0017] A method of operation of a leak detection system using artificial intelligence-based image analysis according to another embodiment of the present invention may include: a step in which at least one camera captures image data; a step in which an image analysis device analyzes the image data to extract feature information for determining whether there is a leak; and a step in which the image analysis device determines whether a leak has occurred by comparing the feature information with a pre-trained model. Effects of the invention

[0019] According to one embodiment of the present invention, image data captured through at least one camera is analyzed to extract feature information for determining whether there is a leak, and the feature information is compared with a pre-trained model to determine whether a leak has occurred. Thus, there is an advantage in being able to determine whether there is a leak based on image data without a separate sensor.

[0020] In addition, according to the present invention, by analyzing temporal changes in feature information extracted from image data to distinguish between changes that occur temporarily and changes that persist for a certain period of time or longer, and by determining whether leakage has occurred based on the persistent changes, there is an advantage in being able to distinguish between temporary changes caused by environmental changes and actual leakage.

[0021] In addition, according to the present invention, since the change component caused by at least one of lighting change, shadow change, and reflected light change is corrected or removed for at least one of color change, brightness change, and shape change included in feature information extracted from image data, there is an advantage of being able to determine whether leakage has occurred even under various environmental conditions.

[0022] In addition, according to the present invention, the location of an area determined to have leaked is calculated as coordinate information within the image data, and an event image including an area corresponding to the coordinate information is generated and transmitted to at least one of an administrator terminal and an external server, thereby providing the advantage of being able to provide information to the outside that can verify the location of the leaked area.

[0023] It should be understood that the effects of this specification are not limited to the matters described above and can be extended to various contents that can be derived from the detailed description of the embodiments of the invention below. Brief explanation of the drawing

[0025] FIG. 1 is a diagram illustrating the overall configuration of a leak detection system (100) using artificial intelligence-based image analysis according to one embodiment of the present invention. FIG. 2 is a drawing illustrating the detailed configuration of an image analysis device (120) according to one embodiment of the present invention. FIG. 3 is a flowchart illustrating a method for determining whether leakage has occurred by analyzing temporal changes in feature information extracted from image data according to an embodiment of the present invention. FIG. 4 is a flowchart illustrating a method for determining whether leakage has occurred after correcting or removing change components included in feature information extracted from image data according to an embodiment of the present invention. FIG. 5 is a flowchart illustrating a method for calculating the location of an area determined to have leaked fluid as coordinate information and generating an event image to transmit to the outside, according to an embodiment of the present invention. FIG. 6 is an example screen diagram of an administrator terminal displaying a plurality of camera images, leak detection results, and event information according to an embodiment of the present invention. FIG. 7 is an example screen diagram of an administrator terminal for setting camera information and detection sensitivity according to one embodiment of the present invention. FIG. 8 is an example of a screen of an administrator terminal that highlights a camera where a leak was detected according to a leak detection result according to an embodiment of the present invention. FIG. 9 is an example screen of an administrator terminal in which the leak detection status and the location of the leak are displayed in real time according to an embodiment of the present invention. FIG. 10 is an example screen diagram of a manager terminal including a playback control function for an event video according to an embodiment of the present invention. FIG. 11 is an example screen of an administrator terminal that can select a log file according to one embodiment of the present invention. FIG. 12 is a drawing illustrating a sensing element installed in a flange and valve pipe according to one embodiment of the present invention. FIG. 13 is a flowchart illustrating the entire process of an operation method of a leak detection system using artificial intelligence-based image analysis according to an embodiment of the present invention. Specific details for implementing the invention

[0026] In describing the embodiments of this specification, if it is determined that a detailed description of known configurations or functions could obscure the essence of the embodiments of this specification, such detailed description is omitted. Furthermore, parts of the drawings unrelated to the description of the embodiments of this specification have been omitted, and similar parts are denoted by similar reference numerals.

[0027] In the embodiments of this specification, when a component is described as being "connected," "combined," or "joined" with another component, this may include not only a direct connection but also an indirect connection in which another component exists in between. Furthermore, when a component is described as "comprising" or "having" another component, unless specifically stated otherwise, this means that it may include additional components rather than excluding the other component.

[0028] In the embodiments of this specification, terms such as first, second, etc. are used solely for the purpose of distinguishing one component from another component and do not limit the order or importance of the components unless specifically stated otherwise. Accordingly, within the scope of the embodiments of this specification, the first component in an embodiment may be referred to as the second component in another embodiment, and likewise, the second component in an embodiment may be referred to as the first component in another embodiment.

[0029] In the embodiments of this specification, distinct components are intended to clearly explain their respective features and do not imply that the components are necessarily separated. That is, multiple components may be integrated to form a single hardware or software unit, or a single component may be distributed to form multiple hardware or software units. Therefore, such integrated or distributed embodiments are included within the scope of the embodiments of this specification, even if not otherwise mentioned.

[0031] FIG. 1 is a diagram illustrating the overall configuration of a leak detection system (100) using artificial intelligence-based image analysis according to one embodiment of the present invention.

[0032] Referring to FIG. 1, a leak detection system (100) using artificial intelligence-based image analysis according to one embodiment of the present invention may include at least one camera (110) and an image analysis device (120).

[0033] At least one camera (110) can capture video data.

[0034] More specifically, the camera (110) may be installed at various locations within the semiconductor process equipment, including piping, valves, flanges, chemical transfer lines, areas around storage tanks, and the bottom surface of the equipment, and configured to continuously photograph areas where there is a possibility of chemical leakage. The camera (110) may be configured to include an image sensor capable of acquiring high-resolution images, and may be configured to include an image sensor having low-light shooting characteristics so that fine liquid traces can be identified even in low-light environments.

[0035] More specifically, the camera (110) may be configured to include a megapixel-class image sensor, and in one embodiment, may be configured to provide a resolution of about 5 megapixels or more, and may be configured to use a CMOS sensor as the image sensor. For example, a 1 / 2.8-inch CMOS sensor may be used to obtain a clear image with noise suppressed even in a low-light environment. Additionally, the camera (110) may be configured to provide a high-resolution image as a main stream and simultaneously provide a relatively low-resolution image as a sub-stream. For example, the main stream may be configured to be set to a 4K level resolution, and the sub-stream may be set to a 720p level so that the real-time processing load is distributed.

[0036] Additionally, the camera (110) may be configured to include an image compression function for efficient transmission of image data. For example, it may be configured to apply H.264 or H.265 compression, thereby reducing network bandwidth usage while maintaining the image quality necessary for leak detection. Additionally, the camera (110) may be configured to include a pinhole lens or a wide-angle lens, and in one embodiment, it may be configured to have a field of view of about 90 degrees or more. Accordingly, it may be configured so that a wide area can be captured with a single camera, allowing the lower part of the pipe, the inside of the equipment, and the surrounding space to be observed simultaneously.

[0037] Additionally, the camera (110) may be configured to have a miniaturized structure to accommodate various installation environments. For example, the camera body may be formed to have a size of several tens of millimeters so that it can be installed inside narrow facilities or between pipes. Additionally, the camera (110) may be configured to have a structure with waterproof and dustproof performance, and in one embodiment, it may be configured to have an IP66 level protection structure so that it can operate stably even in environments with liquid splashing or high humidity. Accordingly, it may be configured to allow continuous shooting even in environments where there is a possibility of direct exposure to chemicals.

[0038] Additionally, the camera (110) can be configured to operate in various temperature and humidity environments. For example, it can be configured to operate from sub-zero temperatures to high temperatures, and can be configured to include a sealed structure to protect the internal circuitry even in high humidity environments. Accordingly, it can be configured to ensure stable image acquisition even in environments where temperature and humidity changes exist, such as semiconductor process lines.

[0039] Additionally, the camera (110) may be configured to include power supply and communication functions. For example, it may be configured to operate by receiving DC power and to transmit video data to an external device via a network. In this case, each camera may be configured to have unique identification information so that captured video data can be distinguished to correspond to a specific location. Additionally, when multiple cameras (110) are installed simultaneously, the video data collected from each camera may be configured to be processed in parallel.

[0040] Additionally, the camera (110) may be configured so that its installation location is selected to focus on capturing specific surveillance areas. For example, the camera may be configured to be preferentially placed at pipe connections, valve joints, the bottom of a chemical storage container, or sections where a history of leakage exists. Additionally, a camera positioned to capture the bottom surface of the equipment may be configured to detect liquid pooling on the floor, and a camera positioned on the side may be configured to capture the path of liquid flow. A camera installed on the ceiling may be configured to capture a wide area downward to monitor multiple leakage points simultaneously.

[0041] Additionally, the camera (110) is not limited to monitoring a single location but can be configured to monitor multiple locations simultaneously. For example, multiple cameras can be configured to be distributed and installed over the entire length of a semiconductor process line, and each camera can be configured to monitor a wide area simultaneously by capturing different areas. Accordingly, it can be configured to acquire image data not only of the leakage occurring at a specific location but also of the path along which the leakage spreads.

[0042] Additionally, the camera (110) may be configured to generate data including various visual changes caused by liquid leakage within the image. For example, changes in the color of the liquid, changes in surface reflection, droplet formation, liquid flow and diffusion patterns, etc., may be configured to be included in the image data, and this information may be subsequently extracted as feature information and used to determine whether there is a leakage. For example, liquid droplets formed on a metal surface may be configured to be captured to have different brightness or reflection characteristics than the surrounding area, and liquid spread on the floor may be configured to be captured to be expressed with a texture different from the surrounding floor pattern.

[0043] Additionally, the camera (110) can be configured to ensure stable image acquisition in response to changes in the lighting environment. For example, it can be configured so that the image brightness is automatically adjusted according to changes in daytime and nighttime or facility lighting conditions, and the shooting conditions can be adjusted so that image distortion caused by shadows or reflected light is minimized. Accordingly, it can be configured so that changes caused by leakage are consistently reflected in the image data even under various environmental conditions.

[0044] Additionally, the camera (110) can be configured to be advantageous in terms of maintenance. For example, it can be configured to be installed via network connection only, without separate complex wiring, and can be configured to have a small, lightweight structure so that the camera position can be changed or additional installations can be easily performed if necessary. Accordingly, a leak detection system can be configured to be established without modifying existing facilities.

[0046] The image analysis device (120) can analyze image data to extract feature information for determining whether there is a leak, and determine whether there is a leak by comparing it with a pre-trained model based on the feature information.

[0047] More specifically, the image analysis device (120) may be configured to receive image data transmitted from the camera (110) as input and to identify visual changes that occur when a chemical leaks within the image. At this time, the image analysis device (120) may be configured to include a deep learning neural network and may be configured to determine whether there is a leak using a model that has learned the difference between the pipe surface, equipment exterior, and floor condition in a normal state and the surface condition in which a leak has occurred.

[0048] More specifically, the image analysis device (120) may be configured to extract feature information from image data, and the feature information may be configured to include at least one of color change, brightness change, shape change, and surface texture change. For example, if an opaque drug leaks, the color of the outer surface of the pipe may be configured to appear different from the surroundings, and if a transparent drug leaks, the surface reflection pattern or brightness distribution may be configured to be formed differently from the surrounding area. Additionally, if liquid accumulates on the bottom surface, a boundary shape different from the existing bottom pattern may be formed, and this boundary may be configured to be expressed in a form that gradually expands.

[0049] Additionally, the image analysis device (120) may be configured to analyze temporal changes in feature information. For example, if a change in brightness occurs at a specific location and is restored to its original state within a certain period of time, it may be configured to be determined as a temporary change, and if the change persists at the same location or the area of ​​change gradually expands, it may be configured to be determined as a leakage.

[0050] For example, in the case where liquid drips finely from the bottom of the flange, a change that initially appears as a point may expand into a linear or area-like form over time, and if this change persists for a certain period of time or longer, it may be configured to be determined as a leak. Additionally, in the case where liquid spreads on the bottom surface, a pattern that gradually expands from a central area to surrounding areas may be formed, and this expansion pattern may be configured to be used as a primary factor in determining a leak.

[0051] Additionally, the image analysis device (120) may be configured to remove or correct change components caused by lighting changes, shadows, and reflected light. For example, when lighting within the process equipment is turned on or off, the brightness of the entire image may change simultaneously, and in such cases, it may be configured to be removed as it is determined to be an environmental change. Additionally, since shadows generated while a worker moves have a form that moves in a certain direction, changes in position that move over a certain period of time may be configured to be removed.

[0052] For example, if reflected light occurs at a specific angle on a metal surface and a bright spot appears, the system may be configured to determine that the bright spot is reflected light and remove it if its position changes rather than remaining in a fixed state. On the other hand, if the same change in brightness is continuously maintained at a specific location, the system may be configured to maintain it as a change caused by leakage.

[0053] Additionally, the image analysis device (120) may be configured to perform color-based analysis and brightness-based analysis in parallel. For example, if a specific chemical has color, it may be configured so that a change in color is used as a primary judgment factor, and if a colorless chemical leaks, a change in brightness contrast or a change in surface reflection characteristics may be configured so that a change in brightness contrast or a change in surface reflection characteristics is used as a primary judgment factor.

[0054] For example, when a transparent liquid spreads on the floor, the reflectivity of the floor surface may change so that certain areas appear brighter or darker than the surroundings, and if these areas expand while maintaining a certain shape, it may be configured to be determined as a leak.

[0055] Additionally, the image analysis device (120) may be configured to set a specific area within the image so that analysis is performed only on that area. For example, the bottom of the pipe connection, the area around the valve joint, the lower space inside the equipment, etc., may be configured to be designated as monitoring areas, and the sensitivity level may be set differently for each area.

[0056] For example, areas where liquid is likely to fall directly can be configured to have high sensitivity, while areas with frequent worker movement can be configured to have low sensitivity. Accordingly, the system can be configured to reduce unnecessary false detections.

[0057] Additionally, the image analysis device (120) may be configured to classify feature information by comparing it with a pre-trained model. For example, it may be configured to determine that the input feature information is a leak if it is similar to a pattern that appears in a leak state, and to determine that it is not a leak if it is similar to a normal state. At this time, it may be configured to detect even minute changes at the pixel level, so that leaks in the early stages can also be identified.

[0058] For example, if fine droplets formed on the outside of a pipe are repeatedly generated, the system can be configured to repeatedly display a change of the same shape at that location, and this repeating pattern can be determined as a leak.

[0059] Additionally, the image analysis device (120) may be configured to calculate the location of the area determined to have a leak as coordinate information within the image data. For example, it may be configured to detect the boundary of the area determined to have a leak and to calculate the center coordinate of the area or a set of coordinates constituting the area. Accordingly, the leak location may be accurately displayed on the image.

[0060] Additionally, the image analysis device (120) may be configured to generate an event image that includes an area corresponding to coordinate information. For example, an image segment before and after the time of leakage may be extracted and configured to form a single event image, and the image may be configured to include the leakage location in a visually highlighted state. Additionally, the event image may be configured to include time information so that the time of leakage is identified.

[0061] Additionally, the video analysis device (120) may be configured to transmit the generated event video to at least one of an administrator terminal and an external server. For example, it may be configured so that the video is displayed on the administrator terminal along with a notification when a leak occurs, and the video may be stored on an external server so that it can be played or analyzed later. Additionally, it may be configured so that the time of the event, the location of the event, and the related video are recorded together, allowing the administrator to immediately check the situation of the leak remotely.

[0062] Additionally, the image analysis device (120) may be configured to process images received from multiple cameras simultaneously. For example, it may be configured to analyze multiple camera images simultaneously, and may be configured to highlight the corresponding camera image when a leak is detected in a specific camera. Additionally, when images of the same location taken from different angles are analyzed simultaneously, it may be configured to determine that a leak has occurred only when a change appears in the same location in the different images, thereby reducing false detections.

[0063] Additionally, the image analysis device (120) may be configured to include image storage and playback functions. More specifically, the image analysis device (120) may be configured to automatically separate and store image segments before and after the point in time when a leak is determined to have occurred, and the image may be configured to be stored together with time information, occurrence location information, and camera identification information. For example, the image for several seconds immediately before the leak is detected and the image for a certain period after detection may be configured to be stored as a single event unit, thereby allowing for continuous verification from the initial process of the leak to the process of its spread.

[0064] Additionally, the stored video can be configured to be played selectively on the administrator terminal, and the location where leakage occurred can be highlighted during playback. For example, coordinates corresponding to the leakage area can be displayed on the video during playback, and control functions such as playback speed adjustment, pause, and loop playback can be included. Furthermore, multiple event videos can be configured to be sorted according to the time of occurrence and provided in a list format, and the administrator can be configured to select and play a video corresponding to a specific time period or a specific camera.

[0065] For example, if leakage occurs repeatedly at the same location, the system can be configured so that event videos corresponding to that location are viewed continuously, allowing for visual confirmation of whether the leakage pattern is repeating.

[0066] Additionally, the image analysis device (120) may be configured to perform additional analysis on the stored image data. For example, it may be configured to display the accumulated number of events occurring at the same coordinates, and may be configured to highlight an area if the frequency of leakage in that area is relatively high. Additionally, it may be configured to display a record of events occurring over a certain period organized along a time axis, so that a tendency for leakage to be concentrated in a specific time period can be identified.

[0067] Additionally, the image analysis device (120) may be configured to include a computational structure for processing multiple image data simultaneously. More specifically, the image analysis device (120) may be configured to process image data received from multiple cameras in parallel, and may be configured to perform independent analysis operations for each image data. For example, it may be configured to include multiple computational cores or graphics processing units so that each camera image is analyzed simultaneously, thereby enabling simultaneous monitoring of multiple potential leakage points.

[0068] Additionally, the image analysis device (120) may be configured to process multiple image data by setting priorities. For example, camera images corresponding to areas with a high probability of leakage may be configured to be processed with a high priority, while images corresponding to areas with little change may be configured to be analyzed at a relatively low frequency. Accordingly, computational resources may be configured to be efficiently distributed.

[0069] Additionally, the image analysis device (120) may be configured to consider the correlation between different camera images during the process of simultaneously analyzing multiple image data. For example, if there are multiple camera images capturing the same pipe section from different angles, it may be configured to determine a leak only when a change occurring at a specific location appears simultaneously in multiple images, and may be configured to exclude changes occurring in only a single image as false detections.

[0070] Additionally, the image analysis device (120) can be configured to perform analysis without delay even in an environment where multiple image streams are input simultaneously. For example, each image data can be configured to be stored in an independent processing queue and analyzed sequentially, and if real-time processing is required, specific image data can be configured to be processed preferentially. Accordingly, it can be configured so that all areas are monitored without omission, even when leak detection is performed simultaneously over a wide process area.

[0072] FIG. 2 is a drawing illustrating the detailed configuration of an image analysis device (120) according to one embodiment of the present invention.

[0073] Referring to FIG. 2, an image analysis device (120) according to one embodiment of the present invention can perform image data analysis and determine whether leakage has occurred, and may be configured to include a storage unit (121), a communication unit (122), and a processor (123).

[0074] The storage unit (121) may be configured to include a volatile storage medium and a non-volatile storage medium, and may be configured to store a control program required for performing operations of the processor (123), image data collection setting information, feature information extraction setting information, leakage judgment condition information, temporal change analysis condition information, environmental change correction condition information, leakage location calculation rule, event image generation rule and event transmission rule.

[0075] More specifically, the storage unit (121) may be configured to accumulate and store image data captured through the camera (110), feature information extracted from the image, record information of temporal change of feature information, result information of determining leakage occurrence, coordinate information of the location where leakage occurred, event image generation history information, and event transmission history information.

[0076] For example, the storage unit (121) may be configured to record camera identification information, video capture time, color change value in a specific area, brightness change value, shape change information, whether the change persists over time, leakage judgment result, and leakage occurrence coordinate information in chronological order. As another example, if leakage occurs repeatedly at the same location, the change pattern, occurrence time, location coordinates, and event video information corresponding to each event may be stored together, and subsequently, the processor (123) may be configured to perform leakage occurrence pattern analysis.

[0077] Additionally, the storage unit (121) may be configured to store feature information change patterns, temporal change duration patterns, image change patterns due to environmental changes, and repeatedly occurring leak location information to improve the accuracy of leak judgment. For example, it may be configured to store the corresponding change pattern when droplets are formed at regular time intervals in a specific pipe section, and may be configured to record setting information so that if the same pattern is detected thereafter, it is judged as a leak.

[0078] The communication unit (122) may be configured to perform data transmission and reception between the video analysis device (120), the camera (110), and an external device. More specifically, the communication unit (122) may be configured to receive video data by being connected to the camera (110), and may be configured to transmit leak detection results, event video, and notification information by being connected to an administrator terminal or an external server.

[0079] For example, the communication unit (122) may be configured to transmit real-time video data from the camera (110) to the video analysis device (120), and may be configured to transmit the event video and location information to the administrator terminal when it is determined that a leak has occurred. Additionally, the communication unit (122) may be configured to transmit the event video to be stored on an external server, and may be configured so that the stored video can be viewed at a remote location.

[0080] Additionally, the communication unit (122) may be configured to be implemented via wired or wireless communication. For example, communication with the camera (110) may be configured to be performed via an internal network or an industrial communication protocol, and communication with an external server or administrator terminal may be configured to be performed via Ethernet, wireless communication, or a dedicated communication network. As another example, when multiple cameras (110) are installed at different locations, the communication unit (122) may be configured to receive video data from multiple cameras simultaneously.

[0081] The processor (123) is a central computational unit of the image analysis device (120) and may be configured to perform image data analysis and determine whether there is a leak by linking with the storage unit (121) and the communication unit (122). More specifically, the processor (123) may be configured to control the sequential execution of feature information extraction, analysis of temporal changes in feature information, classification of change types, correction of environmental changes, comparison with a learned model, determination of whether a leak has occurred, calculation of leak location coordinates, and generation of an event image.

[0082] For example, the processor (123) may be configured to extract feature information including at least one of color change, brightness change, and shape change from image data, and may be configured to distinguish between changes that occur temporarily and changes that persist for a certain period of time by analyzing the pattern of change over time. Additionally, it may be configured to determine that there is a leak when feature information corresponding to a change that persists for a certain period of time is detected.

[0083] Additionally, the processor (123) may be configured to remove or correct change components caused by lighting changes, shadows, and reflected light occurring within the image data. For example, if brightness changes simultaneously across the entire image area, it may be configured to be removed as an environmental change, and if a change appears continuously only at a specific location, it may be configured to be determined as a leak.

[0084] Additionally, the processor (123) may be configured to classify feature information using a pre-trained model. For example, it may be configured to determine that the input feature information is a leak if it is similar to a change pattern that appears in a leak state, and may be configured to determine that it is not a leak if it is similar to a normal state.

[0085] Additionally, the processor (123) may be configured to calculate the location of the area determined to have leaked as coordinate information within the image data. For example, the boundary of the area where the change occurred may be extracted, and the center coordinates or area coordinates of the area may be calculated.

[0086] Additionally, the processor (123) may be configured to generate an event image that includes an area corresponding to coordinate information. For example, images before and after the time of the leakage may be extracted and formed to constitute an event image, and the location of the leakage may be displayed in the image.

[0087] Additionally, the processor (123) may be configured to control the transmission of the generated event video and leakage determination result to an administrator terminal and an external server via the communication unit (122). For example, it may be configured to send a notification to the administrator terminal when leakage occurs, and may be configured to provide an event video so that the leakage situation can be immediately confirmed.

[0088] Additionally, the processor (123) may be configured to control the leakage determination result, event video generation history, and transmission history to be recorded in the storage unit (121), and may be configured to manage the recorded information so that it can be utilized in equipment inspection or maintenance work.

[0090] FIG. 3 is a flowchart illustrating a method for determining whether leakage has occurred by analyzing temporal changes in feature information extracted from image data according to an embodiment of the present invention.

[0091] Referring to FIG. 3, a method for determining leakage according to one embodiment of the present invention may be performed by a feature information change analysis step (S301), a change type classification step (S302), and a leakage determination step (S303).

[0092] First, in step S301, temporal changes in feature information extracted from image data can be analyzed. More specifically, feature information corresponding to at least one of color change, brightness change, and shape change can be extracted from image data captured through a camera, and a pattern of change over time can be calculated by comparing the differences between multiple frames continuously acquired for the same location. For example, if the brightness value gradually increases or decreases in a specific lower pipe area, the change can be configured to be continuously recorded on the time axis, and if liquid spreads on the floor surface, the area of ​​the change region can be analyzed as expanding over time.

[0093] Next, in step S302, at least one of a temporary change and a change that persists for a certain period of time can be distinguished based on the temporal change of the feature information. More specifically, a change that occurs and disappears within a certain period of time may be configured to be classified as a temporary change, while a change that is maintained at the same location for a certain period of time or whose change area continuously expands may be configured to be classified as a continuous change. For example, a shadow change caused by a worker's movement may be configured to be classified as a temporary change because its location may shift or disappear over time, whereas a case where liquid leaks and accumulates on a surface may be configured to be classified as a continuous change because the change is maintained at the same location.

[0094] Next, in step S303, whether leakage has occurred can be determined based on feature information corresponding to changes that persist for a certain period of time or longer. More specifically, if the feature information classified as a persistent change is similar to a leakage pattern defined in a pre-trained model, it can be configured to determine that leakage has occurred; if the change differs from the leakage pattern, it can be configured to determine that it is a non-leakage state. For example, if a change in color or brightness in a specific area takes the form of spreading in a certain direction, it can be configured to determine that it is leakage, while changes caused by simple noise or momentary reflections can be configured not to determine that it is leakage.

[0096] FIG. 4 is a flowchart illustrating a method for determining whether leakage has occurred after correcting or removing change components included in feature information extracted from image data according to an embodiment of the present invention.

[0097] Referring to FIG. 4, a method for determining leakage according to one embodiment of the present invention may be performed by a feature information extraction step (S401), a change component correction or removal step (S402), and a leakage determination step (S403).

[0098] First, in step S401, feature information can be extracted from image data. More specifically, feature information including at least one of color change, brightness change, and shape change can be extracted from image data captured through a camera, and can be configured to calculate change values ​​in pixel units or area units corresponding to a specific monitoring area. For example, if the color of a specific area on the outer surface of a pipe appears different from the surroundings, the color change can be configured to be extracted as feature information, and if the brightness distribution changes as liquid spreads on the floor surface, the brightness change can be included as feature information.

[0099] Next, in step S402, for at least one of the color change, brightness change, and shape change included in the feature information, a change component caused by at least one of the lighting change, shadow change, and reflected light change may be corrected or removed. More specifically, if brightness changes simultaneously across the entire image area, it may be determined as a lighting change and configured to remove the corresponding change component; and if a shaded area moving in a specific direction is detected, it may be determined as a shadow change and configured to remove it. Additionally, if a strong brightness point occurs localized to a specific location on a metal surface, it may be determined as a change caused by reflected light and configured to be corrected.

[0100] For example, when equipment lighting is turned on or off, the overall brightness of the image may temporarily change, but since such changes are unrelated to leakage, they can be configured to be removed; similarly, shadows generated as an operator moves can also be configured to be removed because they have the characteristic of changing positions. On the other hand, the reflection pattern of the liquid continuously formed at a specific location can be configured to be maintained without being removed.

[0101] Next, in step S403, whether leakage has occurred can be determined based on corrected or removed feature information. More specifically, it can be configured to determine that leakage has occurred if a change in color, brightness, or shape is continuously maintained in a specific area even after a component generated by an environmental change has been removed, and it can be configured to determine that leakage has not occurred if the feature information is not maintained after the change component has been removed.

[0102] For example, an area formed as a liquid spreads on the floor surface can be configured to be judged as a leak because it has a light-dark pattern distinct from the surrounding area and maintains a certain shape, and a temporary change in brightness caused by reflected light disappears after being removed, so it can be configured not to be judged as a leak.

[0104] FIG. 5 is a flowchart illustrating a method for calculating the location of an area determined to have leaked fluid as coordinate information and generating an event image to transmit to the outside, according to an embodiment of the present invention.

[0105] Referring to FIG. 5, a leak treatment method according to one embodiment of the present invention can be performed by a position coordinate calculation step (S501), an event image generation step (S502), and an event image transmission step (S503).

[0106] First, in step S501, the location of the area determined to have a leak can be calculated as coordinate information within the image data. More specifically, a set of pixels corresponding to the area determined to have a leak within the image can be extracted, and the boundary of the area can be set, or the center coordinates or area coordinates can be calculated. For example, if an area formed by liquid spreading on a floor surface has a different brightness or color pattern from the surroundings, the area can be configured to be recognized as a single object, and the location of the object can be configured to be expressed as a coordinate value in the image coordinate system. Additionally, if multiple leak areas exist simultaneously, the coordinate information corresponding to each area can be calculated individually.

[0107] Next, in step S502, an event image containing an area corresponding to the coordinate information may be generated. More specifically, image data for a certain interval before and after the point in time when the leakage occurred may be extracted and formed into a single event image, and the location of the leakage may be visually displayed within the event image. For example, a display area may be formed at a coordinate location corresponding to the leakage area, and the area may be displayed in the form of color highlighting or a border. Additionally, the event image may be configured to include time information of occurrence, so that the point in time when the leakage occurred can be identified.

[0108] In addition, the event video is not limited to a single frame but can be formed to consist of consecutive frames, and can be configured so that the process of leakage occurring and spreading is observed over time. For example, the process of leakage, which initially appears as a small area, gradually expanding can be configured to be continuously expressed within the event video.

[0109] Next, in step S503, the generated event video may be transmitted to at least one of an administrator terminal and an external server. More specifically, if it is determined that a leakage has occurred, the event video may be configured to be transmitted to the administrator terminal, and the event video may be configured to be displayed immediately on the administrator terminal. Additionally, the event video may be configured to be transmitted to and stored on an external server, and the stored video may be configured to be playable or analyzed thereafter.

[0110] For example, in the event of a leak, the administrator terminal can be configured to highlight the relevant camera footage, and simultaneously, the event video can be displayed in a pop-up format. Additionally, the external server can be configured to store the event video along with the location coordinates, time of occurrence, and camera identification information, thereby enabling verification of whether leaks recur at the same location in the future.

[0112] FIG. 6 is an example screen diagram of an administrator terminal displaying a plurality of camera images, leak detection results, and event information according to an embodiment of the present invention.

[0113] Referring to FIG. 6, the screen of the administrator terminal can be configured to display images received from multiple cameras simultaneously, and the images corresponding to each camera can be provided arranged in a grid form. For example, images from multiple cameras installed at different locations within the semiconductor process equipment can be configured to be divided and displayed on a single screen, and the administrator can be configured to check the status of each location simultaneously.

[0114] Additionally, the screen of the administrator terminal may be configured to include a camera list area and to display the operating status of each camera. For example, cameras operating normally may be configured to be displayed as active, and if a leak is detected in a specific camera, that camera may be highlighted.

[0115] In addition, the system can be configured to display the camera footage in a way that distinguishes it from other footage when a leak is detected. For example, the border of the camera footage where the leak occurred can be highlighted with a specific color, and an area corresponding to the location of the leak can be displayed within the footage. Furthermore, the leak detection result or reliability information can be displayed in this area, allowing the administrator to intuitively recognize whether a leak has occurred.

[0116] Additionally, the screen of the administrator terminal may be configured to include an event list area, and the history of leakage occurrences may be configured to be displayed in chronological order. For example, the time at which leakage was detected, the corresponding camera information, and event identification information may be displayed in a list format, and the administrator may be configured to select a specific event to view the video at that point in time.

[0117] Additionally, the screen of the administrator terminal may be configured to include an interface for controlling the leak detection status. For example, it may be configured to provide detection start and detection stop functions, and may be configured to enable or disable detection operations for all cameras based on user input.

[0118] In addition, the screen of the administrator terminal may be configured to include an initial screen, a settings screen, and a login screen. For example, the initial screen may be configured to provide real-time monitoring video, an event list, a date selection function, and a screen configuration setting function, and the user may be configured to set the monitored area and display layout.

[0119] Additionally, the login screen may be configured to provide a user information input function for administrator authentication, and may be configured to allow access to system setting changes or event management functions only after authentication is complete. For example, authentication may be configured to be performed through the input of an ID and password, and setting changes may be restricted in an unauthenticated state.

[0120] In addition, the screen of the administrator terminal can be configured to be integrated with the network environment and security functions. For example, it can be configured to receive video data by connecting to a camera via an internal network, or it can be configured to operate independently without an external network connection. Furthermore, access can be restricted through an authentication device to block access by unauthorized users.

[0121] In addition, the administrator terminal screen can be configured to provide a real-time notification function. For example, if a leak occurs, a warning message can be displayed at the top of the screen, and the corresponding camera footage can be highlighted simultaneously. Furthermore, the event video can be automatically called along with the notification, allowing the administrator to immediately check the situation.

[0123] FIG. 7 is an example screen diagram of an administrator terminal for setting camera information and detection sensitivity according to one embodiment of the present invention.

[0124] Referring to FIG. 7, the screen of the administrator terminal may be configured to input and change setting information for multiple cameras, and may be configured to manage identification information, network information, and detection setting values ​​corresponding to each camera individually. For example, the screen may be configured to display a list of cameras, and when a specific camera is selected, a setting window for that camera may be displayed in a pop-up form.

[0125] In addition, the settings screen can be configured to input name information for each camera, descriptive information indicating the installation location, and network address information for receiving video data. For example, it can be configured to input a streaming address corresponding to each camera, thereby enabling a video data connection between the video analysis device and the camera. Furthermore, it can be configured to display the network connection status, allowing the administrator to intuitively know whether each camera is connected.

[0126] Additionally, the settings screen of the administrator terminal may be configured to include an input interface for adjusting detection sensitivity. For example, detection sensitivity for each camera or each monitoring area may be configured to be set in a numerical form, and the value may be configured to adjust the degree of response to changes in feature information when detecting leakage. Accordingly, the setting may be configured to detect even minute changes, or to be modified to detect only changes above a certain level.

[0127] In addition, the detection sensitivity can be configured to be set differently depending on environmental conditions. For example, the sensitivity can be configured to be low for areas where lighting changes frequently or where a lot of reflected light occurs, and to be high for pipe connections or areas under equipment where there is a high possibility of leakage.

[0128] Additionally, the settings screen of the administrator terminal may be configured to include functions for adjusting the camera's image quality and shooting conditions. For example, settings for the zoom ratio, focus status, or image clarity may be configured to be adjusted, and these setting values ​​may be configured to influence the extraction of feature information within the image data more clearly.

[0129] In addition, the settings screen of the administrator terminal may be configured to include a function for entering memo information about the cameras. For example, information such as the installation location of each camera, monitored equipment, types of chemicals used, or maintenance history may be configured to be recorded in text form, and such information may be provided to enable the administrator to understand the camera settings status.

[0130] In addition, the settings screen of the administrator terminal can be configured to reflect changes in settings immediately. For example, if the sensitivity value changes, the value can be transmitted to the video analysis device so that the leakage detection conditions are changed in real time, and if the network address changes, the camera connection status can be reset.

[0131] Additionally, the settings screen may be configured to include control buttons for applying or canceling setting changes. For example, it may be configured to provide a button to save settings entered by the user and a button to restore existing settings, thereby enabling stable management of camera settings.

[0133] FIG. 8 is an example of a screen of an administrator terminal that highlights a camera where a leak was detected according to a leak detection result according to an embodiment of the present invention.

[0134] Referring to FIG. 8, the screen of the administrator terminal can be configured such that when a leak is detected in a specific camera while multiple camera images are displayed simultaneously, that camera image is highlighted to distinguish it from other images. For example, the border of the image area corresponding to the camera where the leak was detected can be displayed in a specific color, thereby enabling the administrator to immediately identify the location where an anomaly occurred among multiple images.

[0135] Additionally, it can be configured to display a warning message at the top of the screen or in a designated area when a leak is detected. For example, a warning message indicating a leak can be configured to be visually highlighted, allowing an administrator to recognize the abnormal situation without any separate operation.

[0136] In addition, the screen of the administrator terminal may be configured to include a function that magnifies and displays a specific camera image while in real-time monitoring mode. For example, when a user selects or double-clicks a specific camera image, the image may be configured to be displayed in full screen or as a magnified area, thereby allowing for a more detailed identification of the location where the leakage occurred.

[0137] Additionally, the enlarged screen can be configured to display the area where leakage is detected more clearly. For example, it can be configured to apply a display box or highlight at a location corresponding to the leakage area, and can be configured to display the leakage judgment result or reliability information together.

[0138] Additionally, the magnified screen can be configured to return to the original multi-screen. For example, if additional input is performed on the magnified image, it can be configured to switch to the initial screen displaying the full camera feed, and can be configured to allow the administrator to check multiple cameras simultaneously again.

[0139] In addition, the screen of the administrator terminal may be configured to include a function for switching between multiple cameras. For example, it may be configured to immediately switch to another camera image even when zoomed in, thereby enabling sequential verification of leakage situations occurring at multiple locations.

[0140] In addition, the screen of the administrator terminal can be configured to continuously reflect leak detection results while maintaining the real-time monitoring status. For example, it can be configured to additionally highlight the corresponding camera when a new leak is detected, and to display multiple events simultaneously along with previously detected leaks.

[0141] In addition, the screen of the administrator terminal can be configured to support high-resolution video display, allowing for the visual confirmation of even minute changes in leakage. For example, it can be configured to maintain image quality even in magnified images, thereby enabling the identification of small liquid droplets or minute flows.

[0143] FIG. 9 is an example screen of an administrator terminal in which the leak detection status and the location of the leak are displayed in real time according to an embodiment of the present invention.

[0144] Referring to FIG. 9, the screen of the administrator terminal can be configured so that a leak detection status is reflected in real time for each image while multiple camera images are displayed simultaneously. For example, a camera image in a normal state can be configured to remain in a normal display state, and when a leak is detected, the corresponding camera image can be highlighted to distinguish it from other images.

[0145] In addition, if a leak is detected, the system can be configured to visually display the location of the leak within the corresponding video. For example, a display box or area indicator may be formed at a location corresponding to the area where the leak occurred, and the leak detection result or reliability information may be displayed along with that area. Accordingly, the system can be configured to allow the administrator to intuitively verify not only whether a leak has occurred but also its location.

[0146] In addition, the screen of the administrator terminal can be configured to be updated in real time. For example, when video data is continuously input, it can be configured to reflect the leak detection results for each frame, and to continuously display changes in the location or diffusion status of the leak area. Accordingly, the process of the leak expanding after its initial occurrence can be configured to be visually confirmed over time.

[0147] In addition, the screen of the administrator terminal can be configured to display warning information indicating the leak detection status. For example, if a leak is detected, a warning message can be displayed at the top of the screen, allowing the administrator to immediately recognize the abnormal situation through the message.

[0148] In addition, the screen of the administrator terminal can be configured to allow access to setting functions even in full-screen monitoring mode. For example, a menu area can be displayed on one side of the monitoring screen, and the user can be configured to access detection settings, camera settings, or event checking functions through that menu.

[0149] In addition, the administrator terminal screen can be configured to allow setting changes while maintaining the real-time detection status. For example, if the detection sensitivity or specific camera settings are changed, the changes can be configured to be applied immediately, and the leak detection results based on the changed settings can be reflected in real time.

[0150] In addition, the screen of the administrator terminal can be configured to display each event independently when multiple leakage events occur simultaneously. For example, when leakage is detected from different cameras, it can be configured to form individual display areas within each image, and leakage information corresponding to each area can be displayed together.

[0151] In addition, the screen of the administrator terminal may be configured to include a split-screen layout setting function. For example, it may be configured so that the user selects a camera via drag-and-drop and places it at a desired location, and accordingly, the arrangement of cameras on the monitor screen may change according to user settings. Furthermore, the number of screen splits may be configured to be adjusted automatically or manually depending on the number of cameras used, and various screen configurations may be provided according to user convenience.

[0152] In addition, the screen of the administrator terminal can be configured to immediately reflect changes in settings. For example, if the screen splitting structure or camera placement is changed, the changed layout can be applied immediately without interrupting monitoring, thereby allowing the screen configuration to be flexibly adjusted while maintaining real-time monitoring.

[0154] FIG. 10 is an example screen diagram of a manager terminal including a playback control function for an event video according to an embodiment of the present invention.

[0155] Referring to FIG. 10, the screen of the administrator terminal can be configured to display an event video generated at the time the leakage occurred so that it can be played. For example, the screen of the administrator terminal can be configured to display a video corresponding to the selected event in the form of a separate playback window, and the video can be configured to provide the location where the leakage was detected.

[0156] Additionally, the event video playback screen may be configured to include an interface for playback control. For example, it may be configured to provide buttons for playing, pausing, and stopping the video, and the playback status of the video may be controlled according to user input.

[0157] Additionally, the event video playback screen may be configured to include a timeline display for controlling the playback section. For example, it may be configured to display a progress bar corresponding to the total length of the video, and may be configured so that when a user selects a specific location, the video moves to that point in time and plays.

[0158] Additionally, the event video can be configured to include the situation before and after the leakage occurs. For example, it can be configured to include video footage from before the leakage was first detected and video footage from after the leakage was detected, thereby allowing the leakage process to be verified over time.

[0159] Additionally, the event video playback screen can be configured to display leak detection results. For example, the area corresponding to the location of the leak within the video can be highlighted, and the leak judgment result or reliability information can be displayed in that area.

[0160] Additionally, the screen of the administrator terminal can be configured to select and play a specific event among multiple event videos. For example, if an event of a specific time period is selected from the event list, the corresponding event video can be displayed in the playback window, and the administrator can be configured to view multiple events sequentially.

[0161] In addition, the event video playback function can be configured to be provided through an intuitive user interface. For example, it can be configured so that video playback is performed with minimal operation, and the playback status can be visually displayed.

[0163] FIG. 11 is an example screen of an administrator terminal that can select a log file according to one embodiment of the present invention.

[0164] Referring to FIG. 11, the screen of the administrator terminal can be configured to display log files corresponding to the event history generated by the leak detection system in a list format. For example, the screen can be configured to list the time of occurrence, event identification information, detection status, and record information for each log item in row units, and the administrator can be configured to select a specific event among multiple logs.

[0165] Additionally, each log item can be configured to be displayed as selectable. For example, one side of each log item may be configured to include a display element indicating the selection status, and when a user selects a specific log item, that item may be displayed as active.

[0166] Additionally, the screen of the administrator terminal can be configured to check event information corresponding to the selected log item. For example, when a specific log is selected, the screen can be configured to display video or event footage from the time of the leakage occurrence corresponding to that log, thereby allowing the administrator to verify past leakage situations.

[0167] Additionally, log files can be configured to be displayed sorted in chronological order. For example, they can be configured so that recent leakage events are displayed at the top, and administrators can be configured to select and view logs corresponding to specific date or time intervals.

[0168] In addition, the screen of the administrator terminal can be configured to allow continuous viewing of multiple logs. For example, multiple log items can be displayed in a list format, and the administrator can be configured to continuously navigate to previous or subsequent logs through scrolling.

[0169] Additionally, the log file selection feature can be configured to analyze leakage patterns. For example, it can be configured to identify leakage events occurring repeatedly at the same location through the logs, and administrators can be configured to identify leakage trends by continuously checking logs within specific intervals.

[0170] In addition, the log file can be configured to include time information for events, allowing for the identification of leakage events that occurred during specific time periods. For example, leakage events occurring at regular intervals can be configured to be identified in the log, enabling the administrator to estimate the abnormal condition of the equipment.

[0171] In addition, the screen of the administrator terminal can be configured so that the relevant event video is immediately displayed upon log selection. For example, when a log item is selected, the screen can be configured to switch to the playback screen of the corresponding event video, and the administrator can be configured to view both log information and video information together.

[0173] In addition, according to one embodiment, the administrator terminal may be configured to view stored event files without a network connection. For example, event video or log files generated when a leak occurs may be configured to be stored on an internal or external storage medium, and the administrator may be configured to select and view the files even in an offline state. Furthermore, the event files may be configured to be movable via an external storage device, allowing the leak situation to be verified even in a separate analysis environment.

[0174] Additionally, according to one embodiment, the administrator terminal may be configured to include a date selection function. For example, it may be configured to allow a specific date to be selected through a calendar interface displayed on the screen, and may be configured to display leakage occurrence history and event video corresponding to the selected date together. Accordingly, the administrator may be configured to check leakage events that occurred during a specific period in chronological order.

[0175] In addition, according to one embodiment, the administrator terminal may be configured to immediately play the corresponding event upon the operation of selecting an alarm history. For example, when a specific item in the alarm list is double-clicked, the event video corresponding to the alarm may be configured to be displayed on a separate playback screen, and the video at the time of the leakage may be configured to be played automatically.

[0176] In addition, according to one embodiment, the administrator terminal may be configured to display the original video and the event video simultaneously. For example, the original video showing the state before the leakage occurred and the event video showing the state where the leakage was detected may be configured to be output in parallel, and the administrator may be configured to compare the two videos to check the changes before and after the leakage occurred. Additionally, it may be configured to provide play, stop, and pause functions, so that a specific section of the video can be repeatedly checked.

[0177] Additionally, according to one embodiment, the administrator terminal may be configured to distinguish and display areas where leakage is detected by color. For example, the area corresponding to the location where leakage occurred may be configured to be highlighted in a specific color, and if different leakage events occur, different colors may be applied to each event. Accordingly, the administrator may be configured to intuitively distinguish the location and pattern of leakage.

[0178] In addition, according to one embodiment, the administrator terminal may be configured to visually analyze leakage occurrence patterns. For example, leakage areas that occur repeatedly at the same location may be configured to appear in the same color or the same display form, and the administrator may be configured to intuitively identify leakage occurrence trends. In addition, the distribution of locations where leakage has occurred may be configured to be checked on the screen, thereby enabling the identification of abnormal conditions that occur repeatedly in specific sections of the equipment.

[0180] FIG. 12 is a drawing illustrating a sensing element installed in a flange and valve pipe according to one embodiment of the present invention.

[0181] Referring to FIG. 12, according to one embodiment, a sensing element may be attached to a connection part of a pipe within a semiconductor facility, a flange joint, the lower part of a valve, and a location where there is a high possibility of fluid leakage.

[0182] More specifically, the sensing material may be configured to change color upon contact with acidic or alkaline substances, or to maintain a constant color under normal conditions while displaying a distinct color difference upon exposure to specific chemicals. The sensing material may be formed in the form of a tape or a sheet and attached so as to adhere closely to the outer surface of a pipe or the perimeter of a flange joint.

[0183] For example, if a minute leak occurs in a gap of a flange connection or a valve joint, the system can be configured so that the color of the sensor changes locally as the leaked chemical comes into contact with the surface of the sensor. Additionally, if the leakage persists, the system can be configured so that the area of ​​color change gradually expands.

[0184] At this time, at least one camera may be configured to capture an area including the area to which the sensor is attached, and an image analysis device may be configured to extract color changes of the sensor as feature information. More specifically, the image analysis device may be configured to detect changes in color values, color distribution, or specific color patterns of the sensor area.

[0185] For example, if a color different from the existing color is formed in a part of the sensor, the image analysis device may be configured to determine that such change is a reaction caused by leakage. Additionally, it may be configured to determine that there is a high probability of leakage if the color change persists for a certain period of time or if the area of ​​change expands.

[0186] In addition, to distinguish between color changes caused by lighting changes or shadows and color changes caused by chemical reactions of the sensor, the image analysis device may be configured to compare color changes in the sensor area and the surrounding area. For example, if the color changes locally only in the sensor area and the same change does not occur in the surrounding area, it may be configured to determine that the change is caused by a reaction of the sensor.

[0187] In addition, when multiple sensors are installed at different locations, the image analysis device may be configured to individually analyze color changes for each sensor location, and may be configured so that if a color change occurs only at a specific location, that location is identified as the point of leakage.

[0188] In addition, the image analysis device may be configured to distinguish the progression of leakage based on the degree of color change of the sensor. For example, it may be configured to determine an initial leakage state when the area of ​​color change is small, and to determine a continuous leakage state when the area of ​​color change spreads widely.

[0189] Additionally, the video analysis device may be configured to transmit an alarm signal to an administrator terminal when a color change of the sensitive area is detected, and may be configured to highlight the video area containing the sensitive area. For example, the location of the sensitive area where the color change occurred may be highlighted on the administrator terminal screen, or the display area may be distinguished.

[0190] In this way, according to an embodiment using a sensor, a leak can be visually confirmed by using a sensor that changes color in response to a chemical substance, and the corresponding color change is configured to be detected in a camera image, thereby enabling a leak determination to be performed by an image analysis device.

[0192] FIG. 13 is a flowchart illustrating the entire process of an operation method of a leak detection system using artificial intelligence-based image analysis according to an embodiment of the present invention.

[0193] Referring to FIG. 13, the operation method of a leak detection system using artificial intelligence-based image analysis according to one embodiment of the present invention may be performed in the following steps: image transmission and detection start step (S1301), control system operation step (S1302), artificial intelligence-based leak detection step (S1303), detection judgment step (S1304), alarm and image storage step (S1305), and restart step (S1306).

[0194] First, in step S1301, at least one camera may be configured to capture a surveillance target area, including piping, valves, storage tanks, connections, and floor areas within the semiconductor facility, to generate image data. More specifically, each camera may be installed at different positions and angles to capture the same facility from multiple angles, and surveillance may be initiated as the generated image data is transmitted to an image analysis device. Additionally, image transmission may be configured to begin simultaneously when a surveillance start command is input from an administrator terminal.

[0195] Next, in step S1302, the control system may be configured to operate. More specifically, an administrator terminal or an external server may be activated to display the status information, video reception status, and monitoring operation status of each camera, and multiple camera images may be displayed in a split-screen format. Additionally, the real-time monitoring environment for the monitored area may be initialized to form a ready state so that the leak detection results generated in a subsequent step can be displayed.

[0196] Next, in step S1303, the image analysis device may be configured to perform leak detection using a pre-trained artificial intelligence model on the input image data. More specifically, the image analysis device may be configured to extract feature information including color changes, brightness changes, and shape changes from the image data, and the extracted feature information may be input into the artificial intelligence model to identify candidate leak areas. Additionally, the device may be configured to analyze change patterns between image frames to detect the diffusion or flow patterns of the liquid.

[0197] Next, in step S1304, the image analysis device may be configured to determine whether leakage has occurred. More specifically, the temporal change of feature information may be analyzed to distinguish between changes that occur temporarily and changes that persist for a certain period of time or longer, and the device may be configured to determine that leakage has occurred if feature information corresponding to a change that persists for a certain period of time or longer is detected. Conversely, if the change is determined to be caused by a momentary change in lighting or reflected light, the device may be configured to determine that it is not leakage.

[0198] Next, in step S1305, if it is determined that a leak has occurred, an alarm may be triggered and an event video may be saved. More specifically, a warning message may be displayed on the administrator terminal, and a video containing the area where the leak was detected may be generated and saved as an event video. Additionally, the event video may be configured to record coordinate information corresponding to the location of the leak, and the video may be configured to be transmitted to the administrator terminal or an external server.

[0199] Finally, in step S1306, the system may be configured to restart after the alarm action so that monitoring is performed continuously. More specifically, it may be configured to maintain the monitoring state after administrator confirmation or according to automatic settings, and subsequently, new video data may be re-analyzed to repeatedly perform leak detection. Additionally, it may be configured to enable identification of the location of repeated occurrences in conjunction with the history of previously detected leaks.

[0201] Although the present invention has been described in relation to some embodiments, various modifications and changes may be made without departing from the scope of the invention as understood by a person skilled in the art to which the invention pertains. Furthermore, such modifications and changes should be considered to fall within the scope of the claims appended to this specification. Explanation of the symbols

[0203] 100: Leak detection system 110: Camera 120: Image analysis device 121: Storage section 122: Communications Department 123: Processor

Claims

Claim 1 In a leak detection system using artificial intelligence-based video analysis, at least one camera (110) for capturing video data; The system includes an image analysis device (120) that extracts feature information for determining whether there is a leak by analyzing the image data, and determines whether a leak has occurred by comparing the feature information with a pre-trained model. The image analysis device (120) extracts the feature information from the image data, wherein the feature information includes at least one of a color change, a brightness change, a shape change, and a surface texture change. By analyzing the temporal change of the feature information, if a brightness change occurs at a specific location and is restored to its original state within a certain period of time, it is classified as a temporary change; if the change continues at the same location or the area of ​​change gradually expands, it is determined as a leak. When liquid falls finely from the bottom of the flange, the device analyzes whether a change appearing in a dot shape expands in a linear or area shape over time, and when liquid spreads on the bottom surface, whether a pattern is formed that gradually expands from a central area to a surrounding area. For at least one of the color change, brightness change, and shape change included in the feature information, the device removes or corrects the change component caused by at least one of a lighting change, a shadow change, and a reflected light change. The device performs color-based analysis and brightness-based analysis on the feature information, wherein a chemical having color In the case of leakage, the above-mentioned color change is used, and in the case of leakage of a colorless chemical, a change in brightness contrast or a change in surface reflection characteristics is used, and a specific area within the image is set and analysis is performed only on said specific area, wherein said specific area includes at least one of the bottom of the pipe connection, the area around the valve joint, and the lower space inside the equipment, and the sensitivity level is set differently for each said specific area.A leak detection system that compares the above feature information with the above-mentioned pre-trained model, determines that there is a leak if the above feature information is similar to a pattern appearing in a leak state, and determines that there is no leak if it is similar to a normal state, calculates the location of the area determined to be leaked as coordinate information within the above-mentioned image data, generates an event video including an area corresponding to the above-mentioned coordinate information, wherein the event video includes an image segment before and after the time of the leak occurrence, includes a state in which the leak location is visually highlighted, and includes time information, transmits the generated event video to at least one of an administrator terminal and an external server, processes multiple image data received from multiple cameras simultaneously, and, when there are multiple camera images capturing the same pipe section from different angles, determines that there is a leak only when a change occurring at a specific location appears simultaneously in multiple images, and excludes changes occurring only in a single image as false detection. Claim 2 delete Claim 3 delete Claim 4 delete Claim 5 A method of operation of a leak detection system using artificial intelligence-based image analysis, comprising: a step in which at least one camera captures image data; a step in which an image analysis device extracts feature information from the image data to determine whether there is a leak, wherein the feature information includes at least one of a color change, a brightness change, a shape change, and a surface texture change; a step in which the image analysis device analyzes the temporal change of the feature information and classifies it as a temporary change if the brightness change occurs at a specific location and is restored to its original state within a certain period of time, and determines it as a leak if the change persists at the same location or the area of ​​change gradually expands; a step in which the image analysis device analyzes whether a change appearing in a point form when liquid falls finely from the bottom of a flange expands in a linear or area form over time, and whether a pattern is formed in which the liquid spreads on the floor surface and gradually expands from a central area to a surrounding area; a step in which the image analysis device removes or corrects a change component caused by at least one of a lighting change, a shadow change, and a reflected light change for at least one of the color change, the brightness change, and the shape change included in the feature information; and a step in which the image analysis device performs color-based analysis and brightness-based analysis on the feature information, wherein the color A step of utilizing the color change in the event of chemical leakage, and utilizing a change in brightness contrast or a change in surface reflection characteristics in the event of colorless chemical leakage; a step in which the image analysis device sets a specific area within the image and performs analysis only on the specific area, wherein the specific area includes at least one of the bottom of the pipe connection, the area around the valve joint, and the lower space inside the equipment, and sets a different sensitivity level for each specific area;A method of operation of a leak detection system comprising: a step in which the image analysis device compares the feature information with a pre-trained model and determines that there is a leak if the feature information is similar to a pattern appearing in a leak state, and determines that there is no leak if it is similar to a normal state; a step in which the image analysis device calculates the location of the area determined to be leaked as coordinate information within the image data; a step in which the image analysis device generates an event image including an area corresponding to the coordinate information, wherein the event image includes an image segment before and after the time of the leak occurrence, includes a state in which the leak location is visually highlighted, and includes time information; a step in which the image analysis device transmits the generated event image to at least one of an administrator terminal and an external server; and a step in which the image analysis device simultaneously processes multiple image data received from multiple cameras, and when multiple camera images capturing the same pipe section from different angles exist, determines that there is a leak only when a change occurring at a specific location appears simultaneously in multiple images, and excludes a change occurring only in a single image as a false detection.

Images