Smoke detection system, smoke detection method and program

Abstract

To suppress excessive detection of smoke in a waste storage pit.SOLUTION: A smoke detection system for detecting combustion smoke generated from a combustion material in a waste storage pit comprises: an imaging section for imaging a top face of a stored waste; a determination section for determining the presence / absence of an image portion of smoke in a picked-up image which is picked up by the imaging section; and an image processing section which generates a detection frame at a position of the image portion of the smoke in accordance with a determination result and generates a mask frame covering the generated detection frame. The determination section inputs the picked-up image to a learned model and performs a determination according to a certainty parameter indicating certainty of the presence of the outputted image portion of the smoke. The image processing section generates the detection frame in the image portion of the smoke in the picked-up image according to the determination and generates the mask frame for masking a predetermined region covering the detection frame. The determination section determines whether or not the smoke in the image portion of the smoke is the combustion smoke based on a relation between the mask frame and the detection frame and detects the combustion smoke.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present invention relates to a smoke detection system, a smoke detection method, and a program for detecting combustion smoke emitted from combustibles in a waste storage pit.

Background Art

[0002] In recent years, in a waste storage pit provided in a waste intermediate treatment plant for storing combustible waste, combustibles in a stored state may be mixed with combusted materials. In particular, when large waste is crushed for pretreatment and then transported to and stored in a waste storage pit, during crushing, a part of the waste consisting of metal and a crushing part consisting of metal of a crusher come into contact with each other, generating sparks, thereby causing a part of the waste to burn and be transported and stored in a combusted state. In particular, when a lithium-ion battery is mixed in the waste, the lithium-ion battery may be impacted by a bucket for stirring the waste and catch fire. Such combusted materials, even if initially in a smoldering state, may catch fire with a large number of combustibles in the surrounding area during storage, and there is a risk that the fire will spread rapidly.

[0003] Therefore, the applicant proposed the technology of the smoke detection system described in Patent Document 1. The smoke detection system described in Patent Document 1 is a smoke detection system for detecting combustion smoke emitted from combustibles mixed with waste in a waste storage pit, in which a determination unit inputs an image captured by an imaging unit into a learned model, and determines whether combustion smoke exists based on a probability parameter indicating the accuracy of the existence of an image portion of the combustion smoke output. Here, the learned model described in Patent Document 1 is a learned model generated in advance by machine learning using, as teacher data, an input / output data set in which teacher images including an image of the upper surface of waste in a stored state in a waste storage pit, an image having an image portion of combustion smoke and an image having no image portion of combustion smoke are input data, and determination results indicating whether an image portion of combustion smoke exists in each of the teacher images are output parameters.

[0004] In the smoke detection system described in Patent Document 1, by focusing on the rising smoke in the initial stage, smoke is detected using a learned model generated by deep learning with an image obtained by capturing a smoke image as teacher data. As a result, it is possible to issue an alarm at the initial stage of the occurrence of a fire and notify a user such as a worker of the smoke generation information.

Prior Art Document

Patent Document

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] According to the smoke detection system described in Patent Document 1 mentioned above, it is possible to capture the smoke in the initial stage of the occurrence of a fire. However, in the video captured by the imaging unit, there may be cases where dirt or waste in the stationary waste storage pit that does not move is similar to smoke, and there is a possibility of over-detection where detection is performed excessively. Therefore, the applicant has found the need to suppress over-detection of smoke in the waste storage pit and has found the need to develop a technology capable of suppressing over-detection of smoke.

[0007] The present invention has been made in view of the above, and an object thereof is to provide a smoke detection system, a smoke detection method, and a program capable of suppressing over-detection of smoke in a waste storage pit.

Means for Solving the Problems

[0008] In order to solve the above-described problems and achieve the object, a smoke detection system according to an aspect of the present invention is a smoke detection system that detects combustion smoke emitted from combustion products mixed with the waste in a waste storage pit where combustible waste is stored, the smoke detection system including: an imaging unit that images the upper surface of the waste stored in the waste storage pit; a determination unit that determines whether or not there is an image portion of smoke in the captured image captured by the imaging unit; and an image processing unit that can generate a detection frame at the position of the image portion of the smoke according to the determination result by the determination unit and can generate a mask frame that covers the generated detection frame, wherein the determination unit inputs the captured image by the imaging unit into a learned model, and performs the determination based on an accuracy parameter indicating the accuracy of the presence of the image portion of the smoke output from the learned model, and the image processing unit generates a mask frame that masks a predetermined region covering the detection frame after generating the detection frame for the image portion of the smoke in the captured image by the determination, and the determination unit determines whether or not the smoke in the image portion of the smoke is the combustion smoke based on the relationship between the mask frame and the detection frame, and detects the combustion smoke.

[0009] In the smoke detection system according to an aspect of the present invention, in the above invention, the learned model is a learned model generated in advance by machine learning using, as teacher data, an input / output data set in which teacher images including an image of the upper surface of the waste in a stored state in a waste storage pit, an image having an image portion of combustion smoke and an image having no image portion of combustion smoke are input data, and determination results as to whether or not there is an image portion of combustion smoke in each of the teacher images are output parameters.

[0010] In the smoke detection system according to an aspect of the present invention, in the above invention, when the detection frame newly generated by the image processing unit is within the region of the mask frame that has already been generated, the image processing unit maintains the already generated mask frame without newly generating a mask frame for the region covering the newly generated detection frame, and the determination unit determines that the smoke in the image portion of the smoke is the combustion smoke when the number of detection frames generated within the range of the mask frame exceeds a predetermined threshold.

[0011] In the smoke detection system according to one aspect of the present invention, in the above invention, when the detection frame newly generated by the image processing unit is outside the region of the mask frame that has already been generated, the image processing unit newly generates a mask frame that covers the newly generated detection frame, and when the number of mask frames generated by the image processing unit exceeds a predetermined threshold, the determination unit determines that the smoke in the image portion of the smoke is the combustion smoke.

[0012] The smoke detection system according to one aspect of the present invention further includes an output unit configured to output the captured image, the detection frame, and the mask frame.

[0013] A smoke detection method according to one aspect of the present invention is a smoke detection method in which a processor having hardware detects combustion smoke emitted from a combustible material mixed with waste in a waste storage pit in which the combustible waste is stored, the method including: an acquisition step of acquiring a captured image of the upper surface of the waste stored in the waste storage pit and storing the captured image in a storage unit; a determination step of reading out the captured image acquired in the acquisition step from the storage unit, inputting the captured image into a learned model, and determining whether or not there is an image portion of smoke in the captured image based on an accuracy parameter indicating the accuracy of the presence of the image portion of smoke output from the learned model; a detection frame generation step of generating a detection frame at the position of the image portion of the smoke according to the determination result in the determination step; a mask frame generation step of generating a mask frame that masks a predetermined region covering the detection frame with respect to the detection frame generated in the detection frame generation step; and a detection step of determining whether or not the smoke in the image portion of the smoke is the combustion smoke based on the relationship between the mask frame and the detection frame and detecting the combustion smoke.

[0014] The smoke detection method according to one aspect of the present invention is, in the above invention, the learned model is an input data of teacher images including an image of the upper surface of waste in a waste storage pit, an image in which an image portion of combustion smoke exists and an image in which no image portion of combustion smoke exists, and an input / output data set in which a determination result as to whether or not an image portion of combustion smoke exists in each of the teacher images is used as an output parameter is used as teacher data, and is a learned model generated in advance by machine learning.

[0015] The smoke detection method according to one aspect of the present invention is, in the above invention, when the detection frame newly generated in the detection frame generation step is within the region of the mask frame already generated in the mask frame generation step, the already generated mask frame is maintained without newly generating a mask frame for the region covering the newly generated detection frame, and in the detection step, when the number of detection frames generated within the range of the mask frame exceeds a predetermined threshold, it is determined that the smoke in the image portion of the smoke is the combustion smoke.

[0016] The smoke detection method according to one aspect of the present invention is, in the above invention, when the detection frame newly generated in the detection frame generation step is outside the region of the mask frame already generated, in the mask frame generation step, a mask frame covering the newly generated detection frame is newly generated, and in the detection step, when the number of mask frames generated in the mask frame generation step exceeds a predetermined threshold, it is determined that the smoke in the image portion of the smoke is the combustion smoke.

[0017] A program according to one aspect of the present invention is a program that causes a processor having hardware to execute a step of detecting combustion smoke emitted from combustion products mixed with waste in a waste storage pit where combustible waste is stored. The step includes: an acquisition step of acquiring and storing in a storage unit an imaging image obtained by imaging the upper surface of the waste stored in the waste storage pit; a determination step of reading out the imaging image acquired in the acquisition step from the storage unit, inputting it into a learned model, and determining whether or not there is an image portion of smoke in the imaging image based on a probability parameter indicating the probability of the presence of the image portion of smoke output from the learned model; a detection frame generation step of generating a detection frame at the position of the image portion of smoke according to the determination result in the determination step; a mask frame generation step of generating a mask frame that masks a predetermined area covering the detection frame with respect to the detection frame generated in the detection frame generation step; and a detection step of determining whether or not the smoke in the image portion of smoke is the combustion smoke based on the relationship between the mask frame and the detection frame and detecting the combustion smoke.

Advantages of the Invention

[0018] According to the smoke detection system, smoke detection method, and program of the present invention, it is possible to suppress over-detection of smoke in a waste storage pit.

Brief Description of the Drawings

[0019]

Figure 1

Figure 2

Figure 3

Figure 4A

Figure 4B

Figure 5A

Figure 5B

Figure 6A

Figure 6B

Mode for Carrying Out the Invention

[0020] Hereinafter, an embodiment will be described with reference to the drawings. In all the drawings of the following embodiments, the same or corresponding parts are denoted by the same reference numerals. Further, the present invention is not limited by the embodiments described below.

[0021] First, in explaining the embodiment of the present invention, the problems found by the present inventor and the intensive studies conducted to solve the problems will be described.

[0022] The inventor of the present invention implemented the technology described in Patent Document 1. Here, the smoke detection system in the waste storage pit for combustible waste according to the technology described in Patent Document 1 focuses on the smoke that rises at the initial stage of combustion, and after detecting the smoke, it issues an alarm to inform people such as workers of the occurrence information of the smoke in the initial stage of the fire. Here, for the detection of the rising smoke, artificial intelligence (AI) including a learning model generated based on teacher data learned by deep learning is used. The learning model uses, as input parameters, teacher images including an image of the upper surface of the waste in the storage state in the waste storage pit, an image with an image portion of combustion smoke, and an image without it, and uses, as output parameters, the determination results of whether or not an image portion of combustion smoke exists in each of the teacher images. An input / output dataset is generated by machine learning using this as teacher data.

[0023] In such a smoke detection system, it is possible to detect the smoke that rises in the waste storage pit at the initial stage of the fire. However, when the inventor of the present invention tried the technology described in Patent Document 1, there was a concern that there was a possibility of over-detecting dirt in the waste storage pit or the waste itself, which is similar to smoke, as smoke. Therefore, as a method for suppressing over-detection, the inventor of the present invention conducted various experiments, observations of smoke, and intensive studies based on these.

[0024] Here, the inventor of the present invention continued to observe the smoke for a long time regarding the characteristics of the over-detected smoke. As a result, the inventor of the present invention noticed the difference between the characteristics of the dirt in the waste storage pit or the waste itself, which is similar to smoke, and on the other hand, the characteristics of the smoke actually caused by flames or combustion through long-term observation of the smoke. In other words, the inventor came up with the idea of paying attention to the movement of the smoke. As a result, the inventor of the present invention devised a technology that can embody a method for distinguishing smoke caused by flames from other objects or phenomena (hereinafter referred to as pseudo-smoke) similar to smoke.

[0025] That is, the inventor of the present invention focused on the fact that in the smoke detection system according to the above-described prior art, the smoke determination criteria by the AI consisting of the learning model generated by deep learning is a smoke detection frame (hereinafter referred to as the detection frame) based on the agreement with the teacher data. In this regard, the inventor found that when the correct answer rate regarding the smoke detection by the detection frame is improved, along with the improvement of the detection accuracy, a detection frame is also generated for pseudo-smoke other than actual smoke (hereinafter referred to as combustion smoke), which may lead to over-detection.

[0026] Therefore, the inventor of the present invention examined means for suppressing over-detection by embodying the characteristics of smoke caused by flames. As a result, a method was devised in which a determination frame (hereinafter referred to as a mask frame) for newly suppressing over-detection is superimposed on the detection frame generated by the above-described learning model, and masking is performed on the generated smoke detection frame under predetermined conditions. That is, it was devised to perform masking on pseudo-smoke not caused by flames or combustion. Furthermore, the inventor came up with the idea of generating and superimposing a predetermined mask frame on the detection frame generated by detecting the detected smoke or pseudo-smoke, and combining the generation of the detection frame and the generation of the mask frame.

[0027] Specifically, as the embodiment of smoke, a mask frame for masking the detection frame is generated in conjunction with the generation of the detection frame. As a condition for determining whether it is combustion smoke or pseudo-smoke, a threshold is set for the number of mask frames. Furthermore, inside the mask frame, when the number of detection frames exceeds a preset threshold and when the number of mask frames exceeds a preset threshold, it is determined that the smoke for which the detection frame was generated is combustion smoke. As a result, it becomes possible to distinguish between an object with movement like smoke in a flame and an object without movement like pseudo-smoke for the detection target, and a method for suppressing over-detection was devised by clarifying the distinction between combustion smoke caused by flames or combustion and pseudo-smoke not caused by flames or combustion in the smoke determination criteria.

[0028] Specifically, in the present invention, the smoke generated by the combustion of the waste stored in the waste storage pit is the detection target, and the smoke that is a characteristic of the combustion smoke, which rises relatively straight and does not stay in a fixed place, is the detection target. On the other hand, for the detection frame that has detected the pseudo-smoke that stays in a fixed place and does not move, a mask frame according to predetermined conditions is set, and it is determined that it is not combustion smoke, that is, pseudo-smoke, and is excluded from the detection target. Thus, among the targets (combustion smoke or pseudo-smoke) for which the detection frame is generated, according to the detection frame generated by the invention described in Patent Document 1 above and the number of mask frames generated by the present invention, the rising smoke can be embodied. Therefore, in the smoke detection system, over-detection for fixed objects such as wall stains and waste can be suppressed, and countermeasures for suppressing over-detection can be realized. More accurate and high-precision smoke detection and determination are possible, and the performance of the smoke detection system can be improved. The following embodiment is devised based on the intensive study of the present inventors as described above.

[0029] (Smoke Detection System) FIG. 1 is a schematic diagram of a part of a waste intermediate treatment plant 100 where a smoke detection system 10 according to an embodiment is provided. The waste intermediate treatment plant 100 includes a waste storage pit 110 and an incinerator section 120.

[0030] (Waste Storage Pit) The waste storage pit 110 includes an entrance / exit 111, a platform 112, a waste storage pit 113, a ceiling 114, a crane mechanism 115, a forced draft fan 130, and a cooling draft fan 131. The forced draft fan 130 and the cooling draft fan 131 are fans that blow air into the incinerator 122 of the incinerator section 120 described later. Note that the forced draft fan 130 is also called an FDF (Forced Draft Fan). The cooling draft fan 131 is also called a CDF (Cooling Draft Fan).

[0031] The entrance / exit 111 is where a waste collection vehicle V that collects and transports combustible waste enters and exits. The platform 112 is an area where the waste collection vehicle V stops when it throws the waste transported into the waste storage pit 113.

[0032] The waste storage pit 113 is provided as a region surrounded by a bottom 113a, side walls 113b and 113c that are erected so as to be substantially orthogonal to the bottom 113a and face each other, and two side walls (not shown) that are erected from the bottom 113a so as to be substantially orthogonal to the side walls 113b, 113c and the bottom 113a and face each other. An entrance / exit 111 and a platform 112 are provided on the side wall 113b. The side wall 113c is a partition wall with the side of the incinerator section 120 located on the incinerator section 120 side with respect to the side wall 113b. An air intake port 113ca of a push blower 130 and a furnace cooling blower 131 is provided at the upper part of the side wall 113c. Waste G is stored in the waste storage pit 113. An upper surface G1 is defined for the waste G in the storage state in the waste storage pit 113.

[0033] The ceiling 114 is provided to face the bottom 113a. In the vicinity of the ceiling 114, the waste storage pit 110 communicates with the incinerator section 120. Gas is sucked from the air intake port 113ca by the push blower 130 and the furnace cooling blower 131, and exhausted through exhaust ports 132 and 133 provided in the incinerator section 120, and is sucked to the incinerator section 120 side. Thereby, the waste storage pit 113 is maintained at a negative pressure, and leakage of odor to the outside from the entrance / exit 111 can be prevented. Arrows A1, A2, A3, A4, A5, A6, and A7 indicate the flow of air in the waste storage pit 110. The flow of air includes a flow that flows in from the entrance / exit 111, a flow that proceeds to the bottom 113a of the waste storage pit 113, a flow that rises from the entrance / exit 111 along the side wall 113b to the ceiling 114, a flow that rises from the upper surface G1 along the side wall 113c to the ceiling 114, a flow that proceeds along the ceiling 114 and flows into the incinerator section 120, a flow that flows into the air intake port 113ca, and the like.

[0034] The crane mechanism 115 is provided in the vicinity of the ceiling 114. The crane mechanism 115 includes a gantry 115a, a trolley 115b, a rope 115c, and a bucket 115d.

[0035] The girder 115a is spanned near the ceiling 114. The trolley 115b is configured to be able to traverse the girder 115a and to be able to wind up and unwind the rope 115c. The rope 115c is configured to include, for example, a support rope for supporting the bucket 115d and an opening and closing rope for opening and closing the bucket 115d. The bucket 115d is provided with a plurality of claws, and it is possible to grip, release, and drop the waste by opening and closing the claws with the opening and closing rope. Note that the bucket 115d may be configured not to use an opening and closing rope such as a hydraulic type.

[0036] The crane mechanism 115 is used for stirring the waste G by the bucket 115d and transporting the waste G to the incinerator section 120. Stirring the waste G is an operation of grasping a part of the waste G existing in a certain place with the bucket 115d and moving it to another place. By stirring, when the waste G contains a plurality of different wastes, a state in which the same type of waste is unevenly distributed can be made more uniform. When stirring the waste G or transporting the waste G, dust D may fly above the upper surface G1 of the waste G. The bucket 115d moves frequently for stirring and the like, and the flying dust D is disturbed by the turbulent flow generated by the stirring.

[0037] Also, when the combustible B mixed with the waste G exists as a fire source, when the combustible B is smoldering and not ignited or the fire has not spread around, the combustion smoke S emitted from the combustible B rises relatively straight along the air flow. In particular, when the combustible B is buried in the waste G, the combustion smoke S rises from between the wastes covering the combustible B, so it is less affected by the turbulent flow generated by the stirring and rises relatively straight.

[0038] (Incinerator section) The incinerator section 120 includes a charging hopper 121, an incinerator 122, and an opening 123. The charging hopper 121 is the part where the waste grabbed and conveyed by the bucket 115d of the crane mechanism 115 is charged. The incinerator 122 is the part that incinerates the waste. The opening 123 is the part where the gas etc. inside the incinerator 122 is discharged, and is connected to a boiler, for example.

[0039] (Smoke detection system) The smoke detection system 10 includes a control device 1 and a camera section 2. The camera section 2 is installed in the waste storage pit 110.

[0040] The camera section 2 is an example of an imaging section, and includes a camera that continuously or at a predetermined frame rate captures an image of the upper surface G1 of the waste G in the storage state in the waste storage pit 113. The camera section 2 is preferably installed so as to be able to capture an overall image over the entire upper surface G1. Also, in order to capture an overall image over the entire upper surface G1, the camera section 2 may be configured to include one camera or a plurality of cameras. When the camera section 2 includes a plurality of cameras, each of the plurality of cameras may be provided at a location spaced apart from each other. The position of the camera section 2 shown in FIG. 1 does not limit the arrangement of the cameras regardless of the number of cameras.

[0041] The camera section 2 has a communication function and transmits the data of the captured image to the control device 1. The camera section 2 includes, for example, a digital camera or an analog camera. The digital camera is, for example, a network camera or an IP (Internet Protocol) camera. The analog camera is provided with, for example, an A / D converter, and transmits the captured image as digital data at a predetermined frame rate to the control device 1. The frame rate is, for example, 60 fps, but is not particularly limited.

[0042] The camera unit 2 is communicably connected to the control device 1 via a network (not shown) composed of, for example, a dedicated line, a public communication network such as the Internet, for example, a local area network (LAN), a wide area network (WAN), and one or a combination of a telephone communication network such as a mobile phone, a public line, a virtual private network (VPN), etc.

[0043] FIG. 2 is a block diagram showing the configuration of the control device 1. The control device 1 includes a control unit 11, an input / output unit 12, a communication unit 13, and a storage unit 14. The control unit 11 includes a determination unit 11a and an image processing unit 11b.

[0044] Specifically, as hardware, the control unit 11 includes a processor such as a central processing unit (CPU), a digital signal processor (DSP), a field-programmable gate array (FPGA), a graphics processing unit (GPU), etc., and a main storage unit (main memory) such as a random access memory (RAM) and a read only memory (ROM). The control unit 11 executes the functions of the determination unit 11a and the image processing unit 11b by executing a program stored in the storage unit 14.

[0045] First, the input / output unit 12 as the output means is configured to be able to notify external devices of predetermined information according to the control by the control unit 11. For example, it includes a display monitor composed of an organic EL (Electro Luminescence) panel, a liquid crystal display panel, etc., and a speaker capable of outputting sound. It displays characters, graphics, etc. on the screen of the touch panel display and outputs sound from the speaker. The input / output unit 12 as the input means is configured using a user interface such as a keyboard, input buttons, levers, a touch panel for manual input provided superimposed on a display such as a liquid crystal, or a microphone for voice recognition. By operating the input / output unit 12 by a user or the like, it is configured to be able to input predetermined information to the control unit 11. That is, the input / output unit 12 is composed of, for example, a touch panel keyboard incorporated inside a keyboard or a display unit to detect touch operations on the display panel, or a voice input device enabling communication with the outside. Note that the input / output unit 12 may be configured with the output unit and the input unit as separate bodies.

[0046] The communication unit 13 is configured to be able to communicate with the camera unit 2. For example, it includes a LAN (Local Area Network) interface board and a wireless communication circuit for wireless communication. The LAN interface board and the wireless communication circuit can be connected to a network (not shown) such as the Internet which is a public communication network. The communication unit 13 is configured to be able to communicate with other devices and servers by connecting to the network.

[0047] The storage unit 14 is composed of a storage medium selected from a volatile memory such as a RAM, a non-volatile memory such as a ROM, an EPROM (Erasable Programmable ROM), a hard disk drive (HDD), and a removable medium. Note that the main memory may be included in the storage unit 14. Further, the removable medium is, for example, a USB (Universal Serial Bus) memory or a disk recording medium such as a CD (Compact Disc), a DVD (Digital Versatile Disc), or a BD (Blu-ray (registered trademark) Disc). Also, the storage unit 14 may be configured using a computer-readable recording medium such as a memory card that can be externally attached.

[0048] The storage unit 14 can store an operating system (OS) for executing the operation of the control device 1, various programs, various tables, various databases, etc. Here, the various programs include an information processing program for realizing processing based on models such as the learning model and the learned model according to the present embodiment. These various programs can also be recorded on a computer-readable recording medium such as a hard disk, a flash memory, a CD-ROM, a DVD-ROM, or a flexible disk and widely distributed. Note that the storage unit 14 may be provided in another server that can communicate via various networks.

[0049] Specifically, the storage unit 14 can temporarily or persistently store the data of the images captured by the camera unit 2. Further, the storage unit 14 can store the data of the images captured by the camera unit 2 as video data. In the storage unit 14, there is stored a program for realizing the determination unit 11a that executes the smoke detection method according to the present embodiment, and the image processing unit 11b that executes the image processing related to the generation of the detection frame and the mask frame. Specifically, in the storage unit 14, there are stored a learned model 14a that is a program executed by the determination unit 11a and is a learned model, and a frame generation program 14b that is a program executed by the image processing unit 11b.

[0050] The control device 1 loads the program stored in the storage unit 14 into the working area of the main storage unit and executes it, and by controlling each component and the like through the execution of the program, a function that meets a predetermined purpose can be realized. That is, by executing the program by the control device 1, the functions of the control device 1 are executed in cooperation with the hardware.

[0051] (Method for generating a learned model) Here, a method for generating the learned model 14a stored in the storage unit 14 will be described. The learned model 14a is generated in advance by machine learning. The machine learning is, for example, deep learning, and may use a neural network such as a convolutional neural network (CNN). At this time, as teacher data, an input-output data set is used in which a teacher image including an image of the upper surface G1 of the waste stored in the waste storage pit 113, an image in which an image portion of the combustion smoke S exists and an image in which no image portion of the combustion smoke S exists, and a determination result as to whether or not an image portion of the combustion smoke S exists in each of the teacher images is used as an output parameter.

[0052] Specifically, in the input / output dataset, an image that has an image portion of the combustion smoke S is labeled with a determination result indicating that the image portion of the combustion smoke S exists, and an image that does not have an image portion of the combustion smoke S is labeled with a determination result indicating that the image portion of the combustion smoke S does not exist. As such teacher images, images obtained by augmenting the data of the images captured by the camera unit 2 can also be used.

[0053] As described above, the bucket 115d moves frequently for stirring or the like, causing dust D to fly up. Therefore, in any of the teacher images, there may be an image portion of the bucket 115d or the dust D that moves. The image portions of the bucket 115d and the dust D can be noise for the detection of the combustion smoke S.

[0054] When the learned model 14a generated in this way receives an image captured by the camera unit 2 as input, it outputs a probability parameter indicating the probability of the presence of the image portion of the combustion smoke S in the image. The probability parameter takes a value from 0 to 1, for example. When the value is 0, the probability of the presence of the image portion of the combustion smoke S is 0%, and when the value is 1, the probability of the presence of the combustion smoke image portion is 100%.

[0055] (Smoke Detection Determination Method) Therefore, in the control device 1, the determination unit 11a inputs an image captured by the camera unit 2 to the learned model 14a, and determines whether or not an image portion of combustion smoke exists in the image based on the probability parameter output from the learned model 14a. Note that for the determination based on the probability parameter by the determination unit 11a, the technique described in Patent Document 1 can be used.

[0056] That is, for example, when the confidence parameter is within the first range, the determination unit 11a determines that there is no image portion of the combustion smoke S. When the confidence parameter is within a second range different from the first range, the determination unit 11a determines that there may be an image portion of the combustion smoke S. When the confidence parameter is within a third range different from the first and second ranges, the determination unit 11a determines that there is an image portion of the combustion smoke S. For example, when the confidence parameter takes a value from 0 to 1, the first range is 0 or more and less than 0.2, the second range is 0.2 or more and less than 0.8, and the third range is 0.8 or more and 1 or less. However, the first range, the second range, and the third range are not limited to these values and can be set as appropriate. The first range, the second range, and the third range are stored in the storage unit 14, for example.

[0057] Also, in the control device 1, when the program is executed, the input / output unit 12 performs notification according to the determination result by the determination unit 11a. That is, the input / output unit 12 is an example of a notification unit. The notification is performed, for example, by displaying an image or outputting sound. Here, the image to be displayed may be an image of characters, symbols, or figures.

[0058] For example, when the confidence parameter is within the second range, the input / output unit 12 notifies the first warning information by an image or sound, and when the confidence parameter is within the third range, the input / output unit 12 notifies the second warning information by an image or sound. The first warning information is notified in a state where the accuracy of smoke detection is relatively low, and is, for example, the first-stage warning for alerting the operator and is also called a light warning. The second warning information is notified in a state where the accuracy of smoke detection is relatively high, and is, for example, the second-stage warning for giving a higher-level warning to the operator and is also called a heavy warning.

[0059] Next, the smoke detection method executed in the control device 1 will be described. First, in the control device 1, the communication unit 13 acquires the image transmitted from the camera unit 2 and stores it in the storage unit 14. Thereby, an acquisition step of acquiring an image of the upper surface G1 of the waste G in the storage state in the waste storage pit 113 is executed.

[0060] Next, the image acquired in the acquisition step is read from the storage unit 14 and input as input parameters to the learned model 14a. The learned model 14a outputs a probability parameter as output parameters. The determination unit 11a executes a determination step of determining whether or not an image portion of the combustion smoke S exists in the acquired image based on the probability parameter.

[0061] If it is determined in the determination step that the probability parameter is within the first range, the determination unit 11a determines that there is no image portion of the combustion smoke S, and the next image is acquired.

[0062] If it is determined in the determination step that the probability parameter is within the second range, the determination unit 11a determines that there may be an image portion of the combustion smoke S, and the standby step is executed. In the standby step, the control device 1 enters a standby mode in which the process waits for a predetermined standby time. The standby mode is executed when the probability parameter is within the second range and there may be an image portion of the combustion smoke S but there may also be an image portion of the dust D. The input / output unit 12 notifies the first warning information. Subsequently, the notification step is executed. That is, the input / output unit 12 notifies the first warning information.

[0063] According to the findings obtained by the present inventor through experiments and the like, the dust D is disturbed by the turbulent flow and rises, but after a lapse of time, the rise of the dust D subsides. On the other hand, the combustion smoke S rises relatively straight and continuously over time. Due to this difference in properties between the dust D and the combustion smoke S, the standby mode is executed in the situation where the first warning information is notified. The standby time is, for example, in the range of 0.5 seconds to 2 seconds, but is not limited thereto.

[0064] Also, when it is determined in the determination step that the confidence parameter is in the third range, the determination unit 11a determines that there is an image portion of the combustion smoke S and executes the standby step. In the standby step, the control device 1 enters a standby mode in which it waits for a predetermined standby time. The standby time in the standby mode is, for example, in the range of 0.5 seconds to 2 seconds, but is not limited thereto. Note that the standby time when determined to be in the second range and the standby time when determined to be in the third range by the determination unit 11a may be the same time or different times, and these times can be set as appropriate.

[0065] After the determination unit 11a determines in the determination step that there is an image portion of the combustion smoke S and the standby step is executed, the notification step is executed. That is, the input / output unit 12 notifies the second warning information. Here, in the present embodiment, as the second warning information, a detection frame 200 set to surround a detected area with a predetermined size and area is displayed on the display constituting the input / output unit 12 (see FIG. 4A).

[0066] On the other hand, the storage unit 14 records the image transmitted from the camera unit 2 as continuous video. When the fire spreads and becomes a fire, the combustion smoke may fill the waste storage pit 113 and the fire source may not be confirmed. Therefore, by starting recording simultaneously with the notification of the second warning information and recording the generation position of the combustion smoke S, it becomes easy to identify the fire source and fire extinguishing can be quickly executed using it in subsequent fire extinguishing activities.

[0067] On the other hand, the communication unit 13 outputs the second warning information outside the control device 1. The communication unit 13 outputs the second warning information to, for example, a waste treatment plant device (not shown) that controls the entire operation of the waste intermediate treatment plant 100. The waste treatment plant device displays the second warning information on an alarm panel or issues an alarm. An operator who is a user is stationed at the waste treatment plant device, and early detection of the combustion smoke S becomes possible.

[0068] (Smoke detection method) Next, a smoke detection method based on the detection frame output based on the first warning information or the second warning information and the mask frame masking the output detection frame in the present embodiment will be described. FIG. 3 is a flowchart for explaining the smoke detection method by the smoke detection system according to the present embodiment. FIGS. 4A, 4B, 5A, 5B, 6A, and 6B are diagrams each showing an example of a display form in the smoke detection method according to an embodiment of the present invention.

[0069] As shown in FIG. 3, in step ST1, a detection determination process for detecting smoke by determining the first range to the third range by the determination unit 11a of the control unit 11 that has read the above-described learned model 14a is executed. In the detection determination process, when the determination unit 11a determines that no smoke has been detected (step ST1: No), the detection determination process for determining the above-described first range to the third range by step ST1 is repeatedly executed. Here, when the determination unit 11a determines that smoke has been detected, it is possible to adopt either the case where any of the above-described first warning information and second warning information has been output or the case where the above-described second warning information has been output.

[0070] On the other hand, when the determination unit 11a determines that smoke has been detected (step ST1: Yes), as shown in FIG. 4A, the image processing unit 11b of the control unit 11 generates a detection frame 200 at the position of the detected smoke and displays it on the output unit of the input / output unit 12 (detection frame generation step). Note that, on the display which is the output unit of the input / output unit 12, substantially the entire upper surface G1 of the waste storage pit imaged by the camera unit 2 is displayed. Further, the detection frame 200 is not limited to a rectangle and may be circular, elliptical, or polygonal with three or more sides.

[0071] Next, it proceeds to step ST2 shown in FIG. 3, and the determination unit 11a determines whether the detection frame 200 output in step ST1 is the first detection frame. Here, the "first detection frame" means a detection frame newly displayed within a set area in a state where no detection frame is displayed within the set area for a predetermined time selected within a range of, for example, 0.5 seconds to 2.0 seconds at the output unit of the input / output unit 12. When the determination unit 11a determines that the displayed detection frame is the first detection frame (step ST2: Yes), it proceeds to step ST3.

[0072] In step ST3, which is the mask frame generation step, the image processing unit 11b generates a mask frame 210 having a size corresponding to the area surrounding the detection frame 200, and as shown in FIG. 4B, displays the mask frame 210 in the area overlapping the detection frame 200 at the output unit of the input / output unit 12. Here, the mask frame 210 is generated and displayed at substantially the same timing as the generation and display of the detection frame 200. Also, the area of the mask frame 210 is set to be larger than 1 times and less than or equal to n times (n > 1) the area of the detection frame 200. In the present embodiment, it is preferable that the area of the mask frame 210 is set to be 2 times or more and 3 times or less (2 ≤ n ≤ 3) the area of the detection frame 200, but it is not limited thereto.

[0073] Next, it proceeds to step ST4 shown in FIG. 3, and the determination unit 11a determines whether the number of frames of the generated mask frame 210 is within a preset threshold. When the determination unit 11a determines that the number of frames of the mask frame 210 is within the threshold (step ST4: Yes), it returns to step ST1.

[0074] Subsequently, when it returns to step ST1 and smoke is detected through detection determination in the same manner as described above (step ST1: Yes), as shown in FIG. 5A or FIG. 5B, a new detection frame 201 or detection frame 202 is respectively displayed. In this case, it means that the object determined to be smoke by the determination unit 11a occurs at the position of the detection frame 201 or detection frame 202. Here, the area of the detection frame 200 and the areas of the detection frames 201 and 202 may be the same or different from each other.

[0075] Thereafter, the process proceeds to step ST2, and the determination unit 11a determines that new detection frames 201 and 202 that are not the first detection frame 200 are generated by the image processing unit 11b (step ST2: No), and the process proceeds to step ST5 shown in FIG. 3.

[0076] In step ST5, the determination unit 11a determines whether or not the appearance locations of the new detection frames 201 and 202 in which smoke is detected and generated are within the range of the mask frame 210 generated in step ST3. Here, when the determination unit 11a determines that the appearance location of the new detection frame 201 is within the region of the mask frame 210 as shown in FIG. 5A (step ST5: Yes), the process proceeds to step ST6.

[0077] In step ST6, the determination unit 11a determines whether or not the number of detection frames 201 generated within the region of the mask frame 210 is equal to or less than a threshold value. In the example shown in FIG. 5A, a new detection frame 201 has appeared within the mask frame 210 where it has already been generated. When the threshold value is two frames, the number of frames of the detection frames 200 and 201 existing within the mask frame 210 is equal to or less than the threshold value. Here, as the threshold value for the number of frames of the detection frames 200 and 201 that are allowed to be generated within the mask frame 210, it is at least one frame, preferably two frames, but may be three frames or more and is not necessarily limited. In this case, a new mask frame for the newly generated detection frame 201 is not generated, and the detection frames 200 and 201 are displayed within the mask frame 210.

[0078] As shown in FIG. 5A, when the determination unit 11a determines that the number of detection frames 200 and 201 existing within the region of the mask frame 210 is equal to or less than the threshold value (step ST6: Yes), the process returns to step ST1, and the above-described steps ST1 to ST6 are sequentially executed according to the determination.

[0079] On the other hand, when the determination unit 11a determines in step ST5 that the appearance location of the new detection frame 202 is outside the range of the mask frame 210 as shown on the left side of FIG. 5B (step ST5: No), the process proceeds to step ST3.

[0080] In step ST3, the image processing unit 11b generates a mask frame 211 (on the right side of FIG. 5B) having a size corresponding to the area surrounding the detection frame 202 in the same manner as described above, and displays the mask frame 211 in the area superimposed on the detection frame 202. Here, the mask frame 211 is generated and displayed at substantially the same timing as the generation and display of the detection frame 202. Also, the area of the mask frame 211 is set to be larger than 1 times and smaller than n times (n > 1), preferably, for example, 2 times or more and 3 times or less (2 ≤ n ≤ 3), but not limited to, the area of the detection frame 201. Then, the process proceeds to step ST4.

[0081] In step ST4 shown in FIG. 3, the determination unit 11a determines whether the number of the displayed mask frames 210 and 211 is equal to or less than a preset threshold value. When the determination unit 11a determines that the number of the mask frames 210 and 211 is equal to or less than the threshold value (step ST4: Yes), the process returns to step ST1, and the above-described steps ST1 to ST6 are sequentially executed according to the determination.

[0082] On the one hand, when the determination unit 11a determines in step ST4 that the number of mask frames 210 and 211 generated by the execution of steps ST1 to ST3 exceeds the threshold (step ST4: No), the process proceeds to step ST7, which is the combustion smoke detection step. Here, the state shown on the left side of FIG. 6A is a state in which, after the mask frames 210 and 211 are generated up to a predetermined threshold such as two frames, for example, a detection frame 203 is generated outside the range of the mask frames 210 and 211. In this case, no new mask frame exceeding the threshold is generated in the area surrounding the detection frame 203, and the process proceeds to step ST7. That is, in step ST4, the determination unit 11a determines that the number of mask frames 210 and 211 exceeds the threshold. Note that the threshold for the number of mask frames is not necessarily limited to two frames, and may be one frame or three or more frames, and the number of frames that can detect the rising of smoke and suppress over-detection can be appropriately set.

[0083] On the other hand, when the determination unit 11a determines in step ST6 that the number of detection frames 200 generated within the mask frame 210 by the execution of steps ST1, ST2, and ST5 exceeds the threshold (step ST6: No), the process proceeds to step ST7, which is the combustion smoke detection step. Here, the state shown on the left side of FIG. 6B is a state in which the number of detection frames 200, 204, and 205 generated within the range of the mask frame 210 exceeds a predetermined threshold such as two frames, for example. In this case, the process proceeds to step ST7. Note that the threshold for the number of detection frames is not necessarily limited to two frames, and may be one frame or three or more frames, and the number of frames that can detect the rising of smoke and suppress over-detection can be appropriately set.

[0084] In step ST4 and step ST6, the process proceeds to step ST7, and the image processing unit 11b releases the mask frames 210 and 211 to make them non-displayed as shown on the right side of FIGS. 6A and 6B. The state where these mask frames 210 and 211 are released is maintained for a predetermined cycle time T. Subsequently, the process proceeds to step ST8. In step ST8, the control unit 11 maintains the released state of the mask frames 210 and 211 in step ST7 for a predetermined time, for example, about 5 seconds. Thereafter, the process proceeds to step ST9, and an alarm is issued from the input / output unit 12. Thus, the smoke detection process according to the present embodiment is completed.

[0085] When the inventor tried the smoke detection method by the above-described smoke detection system 10 at an existing waste treatment plant, it was confirmed that the number of false detections, which had occurred 2300 times (575 times / week) in one month conventionally, was suppressed to 0 times (0 times / week) in four months.

[0086] According to the smoke detection system according to the present embodiment described above, when it is determined that smoke has occurred from the upper surface G1 of the waste G stored in the waste storage pit 110, the detection frames 200, 201, 202, 203, 204, and 205 are generated by the learned model 14a and displayed on the input / output unit 12. By setting the mask frames 210 and 211 in the area covering the detection frames 200 to 205, since the plurality of intermittently generated and displayed detection frames 200 to 205 follow the movement of the smoke, when the plurality of detection frames 200 to 205 exceed the range of the mask frames 210 and 211, or the number of the mask frames 210 and 211 exceeds the threshold, it can be determined that combustion smoke is rising from the waste G. Therefore, it is possible to significantly suppress false detections due to pseudo-smoke.

[0087] Although the embodiments have been specifically described above, the embodiments are not limited to the above-described embodiments, and various modifications based on the technical idea of the present invention are possible. For example, the numerical values given in the above embodiments are merely examples, and different numerical values may be used as necessary. The present invention is not limited by the description and drawings that form a part of the disclosure of the present invention according to the present embodiment.

[0088] For example, the learned model used to determine whether the accuracy parameter is in the first range or the second or third range, and the learned model used to determine whether the accuracy parameter is in the first or second range or the third range may be different learned models. Also, for example, although an example in which the next image is acquired after the execution of the standby mode has been described, it is also possible to execute the standby mode after the next image is acquired and before the determination.

[0089] For example, although the learned model 14a is stored in the storage unit 14 of the control device 1, it is also possible to store the learned model 14a in the storage unit of a server that can communicate with the control device 1 via a network such as a public circuit network. In this case, the image captured by the camera unit 2 is transmitted to the server via the network and stored in the storage unit. Then, the control unit of the server inputs the image to the learned model 14a and transmits the output accuracy parameter to the control device 1.

[0090] (Recording medium) In the above-described embodiment, a program capable of executing the smoke detection method by the control device 1 can be recorded on a computer-readable recording medium of a computer or other machines and devices (hereinafter referred to as a computer or the like). By causing a computer or the like to read and execute the program of the recording medium, the computer functions as the control device 1. Here, a computer-readable recording medium refers to a non-temporary recording medium that accumulates information such as data and programs by an electrical, magnetic, optical, mechanical, or chemical action and can be read by a computer or the like. Among such recording media, removable ones from a computer or the like include, for example, flexible disks, magneto-optical disks, CD-ROMs, CD-R / Ws, DVDs, BDs, DATs, magnetic tapes, memory cards such as flash memories, etc. Also, fixed recording media in a computer or the like include hard disks, ROMs, etc. Furthermore, an SSD can be used as either a removable recording medium from a computer or the like or a fixed recording medium in a computer or the like.

[0091] Also, the program to be executed by the control device 1 according to an embodiment may be stored on a computer connected to a network such as the Internet and provided by being downloaded via the network.

[0092] (Other Embodiments) Also, in the control device 1 according to an embodiment, the above-described "section" can be read as "circuit" or the like. For example, the communication section can be read as a communication circuit.

[0093] Further effects and modifications can be easily derived by those skilled in the art. The broader aspects of the present invention are not limited to the specific details and representative embodiments presented and described as above. Accordingly, various changes are possible without departing from the spirit or scope of the general inventive concept defined by the appended claims and their equivalents. For example, the numerical values and types of information given in the above-described embodiment are merely examples, and different numerical values and types of information may be used as necessary, and the present invention is not limited by the descriptions and drawings that form part of the disclosure of the present invention according to the above-described embodiment.

[0094] For example, in the above-described embodiment, deep learning using a neural network as an example of machine learning is adopted, but machine learning based on other methods may also be performed. For example, other supervised learning such as support vector machines, decision trees, naive Bayes, k-nearest neighbor methods, etc. may be used. Also, semi-supervised learning may be used instead of supervised learning.

Explanation of Signs

[0095] 1 Control device 2 Camera unit 10 Smoke detection system 11 Control unit 11a Judgment unit 11b Image processing unit 12 Input / output unit 13 Communication unit 14 Storage unit 14a Learned model 14b Frame generation program 100 Intermediate waste treatment plant 110 Waste storage pit 111 Entrance / exit 112 Platform 113 Waste storage pit 113a Bottom 113b, 113c Side walls 113ca Intake port 114 Ceiling 115 Crane mechanism 115a guard 115b trolley 115c rope 115d bucket 120 incinerator section 121 input hopper 122 incinerator 123 opening 130 push-in blower 131 furnace cooling blower 132, 133 exhaust port 200, 201, 202, 203, 204, 205 detection frame 210, 211 mask frame B combustion product D dust G waste G1 upper surface S combustion smoke T cycle time V waste collection vehicle

Claims

1. A smoke detection system for detecting combustion smoke emitted from combustion products mixed with waste in a waste storage pit where combustible waste is stored, comprising: an imaging unit that images the upper surface of the waste stored in the waste storage pit; a determination unit that determines whether or not there is an image portion of smoke in the captured image captured by the imaging unit; an image processing unit that can generate a detection frame at the position of the image portion of the smoke according to the determination result by the determination unit, and can generate a mask frame that covers the generated detection frame; the determination unit inputs the captured image by the imaging unit into a learned model, and makes the determination based on a probability parameter indicating the probability of the existence of the image portion of the smoke, which is output from the learned model; after the image processing unit generates the detection frame for the image portion of the smoke in the captured image by the determination, the image processing unit generates a mask frame that masks a predetermined area covering the detection frame; the determination unit determines whether or not the smoke in the image portion of the smoke is the combustion smoke based on the relationship between the mask frame and the detection frame, and detects the combustion smoke smoke detection system.

2. The learned model is a learned model that is generated in advance by machine learning using, as teacher data, an input / output data set in which teacher images including an image of the upper surface of waste in a storage state in a waste storage pit, an image having an image portion of combustion smoke and an image having no image portion of combustion smoke, are input data, and determination results as to whether or not there is an image portion of combustion smoke in each of the teacher images are output parameters. The smoke detection system according to claim 1.

3. When the detection frame newly generated by the image processing unit is within the area of the mask frame that has already been generated, the image processing unit maintains the already generated mask frame without newly generating a mask frame for the area covering the newly generated detection frame. The determination unit determines that the smoke in the image portion of the smoke is the combustion smoke when the number of detection frames generated within the range of the mask frame exceeds a predetermined threshold. The smoke detection system according to claim 1.

4. When the detection frame newly generated by the image processing unit is outside the area of the mask frame that has already been generated, the image processing unit newly generates a mask frame that covers the newly generated detection frame. When the number of mask frames generated by the image processing unit exceeds a predetermined threshold, the determination unit determines that the smoke in the image portion of the smoke is the combustion smoke. The smoke detection system according to claim 1.

5. The apparatus further includes an output unit configured to output the captured image, the detection frame, and the mask frame. The smoke detection system according to any one of claims 1 to 4.

6. A smoke detection method in which a processor having hardware detects combustion smoke emitted from a combustible material mixed with waste in a waste storage pit where combustible waste is stored, An acquisition step of acquiring a captured image of the upper surface of the waste stored in the waste storage pit and storing it in a storage unit, A determination step of reading out the captured image acquired in the acquisition step from the storage unit, inputting it into a learned model, and determining whether or not there is an image portion of smoke in the captured image based on an accuracy parameter indicating the accuracy of the presence of the image portion of smoke output from the learned model, A detection frame generation step of generating a detection frame at the position of the image portion of the smoke according to the determination result in the determination step, A mask frame generation step of generating a mask frame that masks a predetermined area covering the detection frame with respect to the detection frame generated in the detection frame generation step, A detection step of detecting the combustion smoke by determining whether or not the smoke in the image portion of the smoke is the combustion smoke based on the relationship between the mask frame and the detection frame, Smoke detection method.

7. The learned model is a learned model generated in advance by machine learning using, as teacher data, an input / output data set in which teacher images including an image of the upper surface of waste in a storage state in a waste storage pit, an image having an image portion of combustion smoke and an image having no image portion of combustion smoke, are input data, and determination results as to whether or not there is an image portion of combustion smoke in each of the teacher images are output parameters. The smoke detection method according to claim 6.

8. When the newly generated detection frame in the detection frame generation step is within the area of the mask frame already generated in the mask frame generation step, the already generated mask frame is maintained without newly generating a mask frame for the area covering the newly generated detection frame. In the detection step, when the number of detection frames generated within the range of the mask frame exceeds a predetermined threshold, it is determined that the smoke in the image portion of the smoke is the combustion smoke. The smoke detection method according to claim 6.

9. In the case where the newly generated detection frame in the detection frame generation step is outside the area of the mask frame that has already been generated, in the mask frame generation step, a new mask frame covering the newly generated detection frame is generated. In the detection step, when the number of mask frames generated in the mask frame generation step exceeds a predetermined threshold, it is determined that the smoke in the image portion of the smoke is the combustion smoke. The smoke detection method according to claim 6.

10. A program for causing a processor having hardware to execute a step of detecting combustion smoke emitted from a combustible material mixed with the waste in a waste storage pit where combustible waste is stored. The steps are as follows: An acquisition step of acquiring a captured image obtained by imaging the upper surface of the waste stored in the waste storage pit and storing it in a storage unit. A determination step of reading out the captured image acquired in the acquisition step from the storage unit, inputting it into a learned model, and determining whether or not there is an image portion of smoke in the captured image based on an accuracy parameter indicating the accuracy of the presence of the image portion of smoke output from the learned model. A detection frame generation step of generating a detection frame at the position of the image portion of the smoke according to the determination result in the determination step. A mask frame generation step of generating a mask frame for masking a predetermined area covering the detection frame with respect to the detection frame generated in the detection frame generation step. A detection step of determining whether or not the smoke in the image portion of the smoke is the combustion smoke based on the relationship between the mask frame and the detection frame and detecting the combustion smoke, and includes Program.

Images