Fire and / or event detection system and method of use thereof

Abstract

The present invention provides a fire and / or event detection system. The system includes at least first camera means, configured to operate in a range between 700nm to 15,000nm, and at least second camera means, configured to operate in a range between 380nm to 750nm. The system includes computing means provided in communication with at least the second camera means, wherein the computing means includes a machine learning algorithm arranged to process and classify data obtained from at least the second camera means, and the computing means is arranged to receive and process said data and discern and detect the occurrence ooff aann event within the field of vision of at least the second camera means.

Description

[0001] Fire and / or event detection system and method of use thereof

[0002] The invention to which this application relates is a fire and / or event detection system and a method of using the same. In particular, the invention relates to a fire detection system for use in an industrial environment.

[0003] Fire detection systems exist in a variety of forms and are applied in a variety of environments. One of the most common systems employs one or more infrared (IR) camera systems, which can detect differences in temperature across the environment. Such systems work best in cool or ambient environments where heat may be readily detected against a cooler backdrop in instances where it shouldn’t occur. In industrial environments, such as metalworks, glassworks, or other such “hot” environments, thermal camera systems are also commonly used and are successful in reliably detecting flames or other hot objects (such as a glass gob) . However, this can frequently be problematic given the already high ambient temperature in the environment: it can then become difficult to discern a region of a higher temperature which shouldn’t ordinarily be so high. One solution employed by users of such systems is to “blank out” one or more of the observable regions of the camera system, for example, those regions where the temperature is expected to be constantly high, such as furnaces, moulding machinery and the like. This is achieved by splitting the observable region into zones and setting a higher temperature threshold (for example for prompting an alert), in the region(s) of higher temperature.

[0004] While the above method and system serves to prevent constant alerts arising in the regions of higher temperature and focussing on other areas in that environment, problems do therefore arise: for example, if an accident occurs within the blanked out region such as a stray flame emanating from a furnace, or a glass gob getting stuck or falling from one of the troughs they are designed to slide down after cutting and scooping, and subsequently catching fire, this would not necessarily be picked up by the system given that the temperature will most likely fall within the temperature threshold set in place for the blanked out region. This “blind spot” can therefore have serious consequences. Further, thermal camera systems which are conventionally provided in such environments are inadequate when it comes to detecting any other accidents or faults in the system, which may not necessarily give off an increased heat signature. For example, in glassworks environments it is possible that conveying / delivery systems carrying newly formed glass gobs, which are still hot, may suffer from jamming, or one or more of the gobs may inadvertently fall from the conveyor / delivery system. While in some circumstances, a thermal camera may be able to see and detect the thermal indication of a stray gob having fallen from the conveyor / delivery system, or the flame / fire which may ignite as a result of it, it is possible that the gob could simply fall out of the line of sight of the camera and thus the heat signature may not be detected until the flames have grown substantially, creating more danger.

[0005] Visual cameras may be used to focus on particularly hot areas / regions, but this requires constant monitoring from personnel, which can be costly, and may also still be missed or not spotted in a timely manner if the person monitoring is not fully paying attention or is not viewing the camera feed at certain points in time. Similarly, the visual cameras may be used to monitor the progress of, for example, glass gobs on a conveyor / delivery system. However, if monitoring personnel simply miss, or do not see the gob fall from the trough or other section, or they miss a jam occurring in these or other such automated systems (due to poor resolution of the video feed or not observing the exact location at the exact moment of the event), the situation may worsen quickly before it can draw the attention of monitoring personnel.

[0006] It is therefore an aim of the present invention to provide an improved fire detection system which overcomes the aforementioned problems associated with the prior art.

[0007] It is a further aim of the present invention to provide a method of using an improved fire detection system which overcomes the aforementioned problems associated with the prior art.

[0008] According to a first aspect of the invention there is provided a fire and / or event detection system, said system including: at least first camera means, configured to operate in a range between 700nm to 15,000nm; at least second camera means, configured to operate in a range between 380nm to 750nm; characterized in that the system includes computing means provided in communication with at least the second camera means, said computing means including a machine learning algorithm arranged to process and classify data obtained from at least the second camera means, the computing means arranged to receive and process said data and discern and detect the occurrence of an event within the field of vision of at least the second camera means.

[0009] Typically, said computing means is in communication with both the first and the second camera means. Preferably, data from the first camera means is processed alongside data from the second camera means, to discern and detect the occurrence of an event within the field of vision of the first and second camera means. Typically, said machine learning algorithm is programmed to recognize a standard set of working conditions and / or parameters, and thus detect events and / or occurrences falling outside of said conditions and / or parameters. In some embodiments, the standard set of working conditions and / or parameters are input / programmed into the machine learning algorithm according to the particular setting / environment in which the system is located.

[0010] In some embodiments, said first camera means is configured to operate in a range between 700nm to l ,400nm.

[0011] In other embodiments, said first camera means is configured to operate in a range between 8,000nm to 15,000nm.

[0012] In one embodiment, said first camera means is configured to operate, selectively, in a range between 700nm to l ,400nm, or between 8,000nm to 15,000nm.

[0013] In some embodiments, said first camera means may be configured to operate in a range between 700nm to l ,400nm, and between 8,000nm to 15,000nm.

[0014] Typically, said first camera means is provided to operate as a thermal imaging camera means.

[0015] Typically, said second camera means is provided to operate in the visible spectrum.

[0016] In some embodiments, the second camera means is provided as a lens-less camera means. Typically, a machine learning algorithm is provided which is arranged, in use, to transform light detected by a sensor into a reconstructed image in the field of view of the camera means. Typically, said machine learning algorithm is programmed to divide the field of view of the first and / or second camera means into a grid comprising two or more sections and subsequently analyse each section sequentially, in use. Such an approach ultimately results in faster and more reliable detections by the system. Traditionally, high-resolution images which are to be reviewed / analysed are compressed, reducing the number of pixels in an image and thus reducing the level of detail which is visible. By maintaining the image size / resolution, and dividing it into smaller sections of a grid, each to be analysed individually and sequentially, the level of detail being reviewed by the system is maintained, ensuring that any relatively small events which occur, which may otherwise not be detected in a compressed image, will be detected by the system of the present invention. Given the processing speed which computing means may achieve now, the additional time taken to review extra “image” is minimal, and consequently the speed, reliability and effectiveness of the system is greatly improved.

[0017] Thus, the system according to the present invention enables the detection of events, accidents and / or other such occurrences which may not be detected by conventional IR camera systems initially, or at all. The provision of camera means coupled with computing means incorporating machine learning technology also enables the system to detect, instantaneously, the moment any non-thermal events, as well as thermal events, occur and which allows for an immediate reaction / response.

[0018] In some preferred embodiments, where a thermal event is detected by the first camera means, the machine learning algorithm is programmed to clarify and / or confirm said event using the second camera means, which will analyse / review the event at the specific location in its field of view, in use. Typically, said second camera means is arranged to prioritize and analyse specifically the detected event, enabling subsequent review / analysis, in use. In some embodiments, said second camera means is arranged to zoom in and / or focus on the detected event. In some embodiments, the second camera means may be provided with optical zoom means.

[0019] Preferably, said first camera means includes one or more camera heads.

[0020] Preferably, said second camera means includes one or more vision camera heads. Typically, said second camera means includes one or more machine vision camera heads.

[0021] In some embodiments, one or more camera heads of said first camera means may be paired with one or more camera heads of said second camera means.

[0022] In some embodiments, the system includes a plurality of camera heads associated with the first camera means, arranged to be located in a plurality of positions in the environment being monitored. Typically, a plurality of camera heads associated with the second camera means is provided, each head arranged to be located in a plurality of positions in the environment being monitored. Preferably, each camera head of the first camera means is paired with a camera head of the second camera means.

[0023] In preferred embodiments, the first camera means and the second camera means are provided to be located in a single camera head. Typically, the system may include a plurality of camera heads located in a plurality of positions in the environment being monitored, each camera head comprising first camera means and second camera means therein. In some embodiments, the or each camera head may include further sensing means. Typically, said further sensing means may include temperature sensing means, air flow sensing means.

[0024] In some embodiments, the or each camera head may include air supply, air circulation and / or air blowing means provided therewith. Typically, air blowing means are provided to blow air over one or more lens of the first and / or second camera means, thereby ensuring said lens are kept clean and free of dirt / particulates which may obstruct or reduce the quality of image obtained by said camera means.

[0025] In some embodiments, air circulation means are provided and arranged to circulate air through housing of the or each camera head, in use. In some embodiments, air supply means are provided to supply an external source of air to the or each camera head. Providing an air blowing means at the lens of the camera means ensure that the lens may be kept clean and free of dirt which may otherwise obstruct the field of view or reduce the image quality. The air which is provided to be blown at the lens, may also be used to circulate through housing of the camera head in order to keep the system components at a desired temperature. This is particularly beneficial when located in environments which may have an increased ambient temperature, such as industrial glassworks or metalworks. If the ambient temperature is too hot and in order therefore to avoid circulating already hot air, an external supply of air, at a cooler temperature, may be provided which can circulate through the camera head and on the lens, to reduce the overall temperature of the components of the system.

[0026] In some embodiments, actuation means are provided which are arranged, in use, to activate the supply of air on to one or more lens of the first and / or second camera means, and / or circulate air through a housing of the or each camera head. Typically, said actuation means are provided in communication with temperature and / or air flow sensing means, such that when a predetermined threshold is met, the actuation means activate a supply of air, in use.

[0027] Preferably, at least the first and second camera means of the system are provided for location in an industrial environment. Typically, the first and second camera means are provided for location in an industrial environment such as a metalworks, glassworks, or other such environment having one or more furnaces, glass-moulding machinery or other such heat sources.

[0028] Preferably, the system further includes communication means, arranged to enable communication between at least the first and second camera means, and the computing means.

[0029] Typically, said communication means are provided to communicate live visual and / or thermal data acquired from the first and / or second camera means to the computing means, which is arranged to analyse and detect events and / or occurrences falling outside of a standard set of working conditions and / or parameters. Further typically, the communication means are further arranged to communicate said detections to a remote location and / or user or users.

[0030] In one embodiment, said communication means are provided to be in communication with machinery in the location that is being monitored. Typically, the communication means are arranged to be able to communicate “shut-down”, “slow-down” or other such commands to machinery or a section of machinery for which there has been detected an event and / or occurrence falling outside of a standard set of working conditions and / or parameters. In one embodiment, said communication means are provided to be in communication with safety systems installed in the location. For example, the communication means may be in communication with a fire suppression system installed in the location, and thus may communicate to that system to activate, or activate a certain section, as required.

[0031] In some embodiments, the system further includes display means, provided to relay live visual data from the first and / or second camera means. Typically, the display means may be located on a camera head in which the first and second camera means are located. Additionally or alternatively, display means may be provided in a location remote from said camera head, for example, in a control room. By providing display means on the camera head itself, this can enable a user to see exactly what is being viewed by the first and / or second camera means and physically adjust the camera head such that it is directed exactly where it is required.

[0032] In some embodiments, said display means may further provide thereon information pertaining to the status of the system, for example, temperature of the first and / or second camera means, air flow status, detected events etc.

[0033] In some embodiments, display means are provided in the form of a downloadable application on a mobile device, and said communication means are provided to communicate live visual and / or thermal data acquired from the first and / or second camera means to said display means.

[0034] In some embodiments, the system may further include alert means. Typically, said alert means may be provided to be audio and / or visual alert means. Typically, communication means of the system are in communication with the alert means, and are arranged to initiate an alert when there has been detected an event and / or occurrence falling outside of a standard set of working conditions and / or parameters.

[0035] Typically, the system includes data storage means. Thus, visual and / or thermal data may be stored and replayed / reviewed at a later date, as required by a user.

[0036] In one embodiment, said computing means are arranged to permit images / data from the first camera means to be overlayed and / or paralleled with data from the second camera means.

[0037] Preferably, said computing means includes machine vision software incorporated therewith. Typically, said software includes an artificial neural network model and as data is captured and received by the second camera means, said data is compared with the neural network model and any anomalies / discrepancies etc. may be identified.

[0038] Typically, the neural network model is arranged to be updated as each event or occurrence happens.

[0039] In another aspect of the present invention, there is provided a method of detecting a fire and / or event occurring using a fire and / or event detection system as defined above, said method including the steps of: providing at least first camera means, configured to operate in a range between 700nm to 15,000nm, and at least second camera means, configured to operate in a range between 380nm to 750nm, of the system to observe and monitor a location or environment; receiving and processing data from at least the second camera means and, utilising the machine learning algorithm incorporated into the computing means of the system, processing and classifying the data to discern and detect the occurrence of an event within the field of vision of the at least second camera means.

[0040] Typically, said computing means is in communication with both the first and the second camera means and data from the first camera means is processed alongside data from the second camera means to discern and detect the occurrence of an event within the field of vision of the first and second camera means.

[0041] Typically, the machine learning algorithm is programmed to recognize a standard set of working conditions and / or parameters, and thus detect events and / or occurrences falling outside of said conditions and / or parameters.

[0042] In some embodiments, the standard set of working conditions and / or parameters are input / programmed into the machine learning algorithm according to the particular setting / environment in which the system is located.

[0043] Typically, said machine learning algorithm is programmed to divide the field of view of the first and / or second camera means into a grid comprising two or more sections and subsequently analyse each section sequentially.

[0044] In some preferred embodiments, where a thermal event is detected by the first camera means, the machine learning algorithm is programmed to clarify and / or confirm said event using the second camera means, which will analyse / review the event at the specific location in its field of view.

[0045] In one embodiment, communication means provided with the system, upon detection of an event and / or occurrence falling outside of a standard set of working conditions and / or parameters, will perform one or more of the following actions: communicate the detection of said event and / or occurrence to a remote location and / or user or users, via email, SMS messaging, digital signal, push notifications through a mobile application, and / or other forms of electronic communication; communicate “shut-down”, “slow-down”, “reject product” or other such commands to machinery or a section of machinery for which there has been detected an event and / or occurrence; communicate to a safety system installed in the location or environment to activate, or activate a certain section, as required; and / or communicate with alert means provided associated with the system to initiate an audio and / or visual alert when there has been detected an event and / or occurrence.

[0046] Preferably, the computing means includes machine vision software incorporated therewith, said software including an artificial neural network model, and as data is captured and received by the second camera means, said data is compared with the neural network model and any anomalies / discrepancies etc. are identified.

[0047] Typically, the neural network model is updated as each event or occurrence happens.

[0048] Typically, at least the first and second camera means of the system are located in an industrial environment.

[0049] Embodiments of the present invention will now be described with reference to the accompanying figures, wherein: Figure 1 illustrates a basic schematic of an environment in which a system in accordance with an embodiment of the present invention is installed.

[0050] Referring now to Figure 1 there is shown an industrial environment, which in some examples may be in the form of a glassworks, although other such environments which may have one or more heat sources would be equally appropriate for the installation of a system according to the present invention. Environments without heat sources would also be suitable; however, the present example describes the system employed within a glassworks. In this particular example, molten glass gobs 1 are produced in a glass melting furnace, pumped through pipes, cut to size and subsequently deposited from the piping 3 into a series of troughs 5, which delivers the gobs 1 to a mould 7 for blowing and shaping. The gobs 1 are cooled and subsequently transferred to a conveyor 9, which then transfers the newly formed glass bottles 1 to the next stage. The present invention provides a system to monitor and detect the occurrence of fires or other events which may occur and which fall outside of a preset standard set of conditions and / or parameters, for example, gobs 1’ falling from the troughs 5 or glass bottles 1 ” falling on or off of the conveyor 9, or otherwise not following the prescribed path within the bottle forming machine and / or on the conveyor 9, which could potentially flame-up and start a fire or cause jam- ups or other such problems. Other such events which may be detected by the system of the invention include the malfunctioning and / or misalignment of, for example, machinery of the system such as robotic arms which transfer the gobs 1 from one stage to another, or if any troughs 5 potentially fall out of position / misalign, which can result in gobs 1’ falling. Such misalignments may not be readily viewable by personnel, however, would be instantly detected by the cameras of the present system. Primarily, within the industrial environment, the system includes first camera means in the form of one or more cameras 11 , which are configured to operate in a range between 700nm to 15,000nm, and second camera means in the form of one or more cameras 13, which are configured to operate in a range between 380nm to 750nm. The first camera means may be operable, according to the requirements of the system and environment in which it is located, anywhere from the “red” regions of the visible spectrum to the long-wavelength infrared region, that is to say, between 700nm and 15,000nm. In some embodiments, the camera 11 may be configured to operate in a range between 700nm to l ,400nm (near-infrared, NIR), or alternatively in a range between 8,000nm to 15,000nm (long-wavelength infrared, LWIR). In other embodiments of the present invention, the camera 11 may be configured to operate, selectively, in a range between 700nm to l,400nm (NIR), or between 8,000nm to 15,000nm (LWIR). Essentially, the first camera 11 is provided to act as a thermal imaging camera with a focus on detecting “thermal” events. The cameras may be provided as individual units located in pairs or groups, or separately, or may be provided in a single integral body or camera head. In most embodiments, multiple cameras 11 and 13 (or multiple camera heads each including both cameras 11 , 13, as a twin camera head) will be provided located in various positions around the environment, ensuring that as many sections and pieces of apparatus and machinery within the environment are being monitored. The system includes computing means which are in communication with at least the second cameras 13, and which includes a machine learning algorithm incorporated therewith that is programmed to process and classify data obtained from the second cameras 13, enabling the computing means to receive and process the data, and discern and detect the occurrence of an event within the field of vision of the second cameras 13. In some embodiments, the computing means could be housed inside the camera. In some embodiments of the invention, the second camera 13 can be provided as a lens-less camera, and the machine learning algorithm acts to transform light detected by a sensor of the camera 13 into a reconstructed image in its field of view.

[0051] In preferred embodiments of the invention, the computing means is also in communication with the first cameras 11 , and so data obtained therefrom may be processed alongside and in parallel with the data from the second cameras 13, enabling the system to discern and detect the occurrence of an event within the field of vision of either or both sets of cameras 11 , 13. The machine learning algorithm provided is programmed to recognize a standard set of working conditions and / or parameters. This thus enables it to detect any events or other such occurrences falling outside of the working conditions and / or parameters. The standard conditions may be pre-set on to the algorithm and also tailored to reflect the specific environment into which the system is to be installed. The algorithm can also be continuously updated and taught over time as events or occurrences which are detected can be reviewed and deemed whether or not to be acceptable etc. The present invention therefore enables the detection of events, accidents and / or other such occurrences which may not be detected by conventional IR camera systems initially, or at all. For example, in embodiments where the first cameras 11 are operating in a “thermal imaging” region, that is to say, between 8,000nm and 15,000nm, there would not necessarily be a requirement to set a temperature threshold in a particular zone, or “blank” it out in respect of those cameras 11 , although that would still be possible. Instead, the second cameras 13, which operate in the visible spectrum range, that is to say, between 380nm and 750nm, can monitor the respective regions and instantly detect anomalies, events or occurrences falling outside the pre-set working conditions / parameters. This means that any stray flames or fires which may start in areas of already high temperature will still be detected by the system. Furthermore, the provision of the second cameras 13 coupled with computing means incorporating machine learning technology enables the system to detect, instantaneously, the moment any non-thermal events occur, for example, the falling of molten gobs 1’ from the troughs 5 or bottles 1 ” from the conveyor 9 —this removes the danger of the gob 1’ or bottle 1 ” falling into a position out of sight: the moment the gob 1’ falls from the trough 5 or bottle 1” falls from the conveyor 9 is captured and detected, allowing for an immediate reaction / response.

[0052] A further feature of the system, which enables improved detection of events which may occur, is the provision of the machine learning algorithm being programmed to be able to divide the field of view of the cameras 11, 13 into a grid comprising two or more sections and subsequently analyse each section sequentially. This approach ultimately results in faster and more reliable detections by the system. Traditionally, high-resolution images which are to be reviewed / analysed are compressed, reducing the number of pixels in an image and thus reducing the level of detail which is visible. By maintaining the image size / resolution, and dividing it into smaller, individual sections of a grid, each to be analysed individually and sequentially, the level of detail being reviewed by the system is maintained, ensuring that any events that are small in size or at a distance from the cameras 11 , 13 which occur (for example a stray or falling gob 1’ or bottle 1” or a spark or flame igniting), which may otherwise not be detected in a compressed image, will be detected by the system of the present invention. Given the processing speed which computing means may achieve now, the additional time taken to review each extra “image” is minimal, and consequently the overall speed, reliability and effectiveness of the system to detect issues is greatly improved. This also enables a single camera head to monitor a larger area than would otherwise be possible. In addition, where a thermal event is detected by the first camera 11 , the machine learning algorithm can be programmed to clarify and / or confirm said event using the second camera 13, which will analyse / review the event at the specific location in its field of view, in use. This can be achieved with or without the grid system discussed above being present, and the second camera will prioritize and home in on the specific event for immediate review and analysis. This can be employed even where the grid system is present, but another option, where the grid system is employed, is rather than wait until that grid is reached, in sequential order, for analysis, it will be prioritised and reviewed immediately, again increasing the efficiency and reliability of the system. As an additional option, if the detected thermal event is below a predetermined size or resolution threshold, as well as being prioritized by the second camera 13 as mentioned above, the second camera 13 may be provided with a zoom function such as an optical zoom, enabling it to zoom in and focus on the detected thermal event, without any loss in image quality, enabling the algorithm to review and analyse it sufficiently well to make a determination as to the nature of the event, and whether further action is required to be taken.

[0053] Communication means, wired and / or wireless, are provided with the system, firstly ensuring communication between the first and second cameras 11 , 13, and the computing means. The communication means are further provided to enable communication of the detection of events and / or occurrences falling outside of the standard set of working conditions and / or parameters, such as via email, SMS messaging, push notifications through a mobile application, and / or other forms of electronic communication to a remote location and / or user or users. The system, via the communication means, may also be provided to be able to communicate directly with various forms of machinery or other such automated apparatus within the environment in which the system is installed. Thus, if an event is detected with a specific piece of machinery, or a section thereof, rather than waiting for a response from a human responder to a notification, the system can communicate a “shut-down”, “slow-down” or other such command to the machinery or apparatus. For example, in the illustrated example of Figure 1 , the system may detect the stray gob 1’ which has fallen from the troughs 5, causing a fire risk to break out. The fall may be the result of some misalignment or other malfunction between the piping and the trough 5, and so the system can communicate a “shut-down” or “slow-down” command to prevent any further gobs 1 being deposited along a specific trough, or other such action, at least until a human responder is able to rectify the error. Conventional systems used in the prior art rely solely on thermal IR cameras and so, in the event where a gob 1’ falls from the trough 5 which does not result in a fire, or an initial flame occurs out of sight of any cameras provided, there would be nothing to spot the resultant incident and subsequently command the machine to stop sending more gobs 1 , resulting potentially in even greater damage — the system of the present invention solves this problem.

[0054] Further sensors may be incorporated into the camera heads containing the first and second cameras 11 , 13, for example, temperature and / or air flow sensors. In addition, the camera heads may include air supply, air circulation and / or air blowing means provided therewith. Air blowers are provided to blow air over the lenses of the cameras 11 , 13, acting as an air purge to ensure they are kept clean and free of dirt / particulates which may obstruct or reduce the quality of image obtained by the cameras. Means to circulate air through the camera head housing can be provided to keep the temperature of the internal components at an acceptable level, and in some embodiments of the invention, air supply means can be provided to supply an external source of air to the camera head. The air which is blown at the lenses from the blowers may come from a single source and also be used to be circulated through camera head housing. If the ambient temperature is too hot and in order therefore to avoid circulating already hot air, the external supply of air, at a cooler temperature, can then be provided which can circulate through the camera head and on the lens, to reduce the overall temperature of the components of the system. In some embodiments, the air may be circulated through the housing to provide a cooling effect, however, the air may not be directed to flow directly over the exposed electronics — this will avoid the possibility of moisture which may be contained within the air contacting the electronics and causing damage. Instead, the electronics are cooled via a secondary cooling effect of the heat sinking into the housing body. Actuation means in the form of an air flow switch may be provided to activate the supply of air on to the lenses and / or circulate air through the camera head housing, the actuation means being provided in communication with the temperature and / or air flow sensors, such that when a predetermined threshold is met, the actuation means activate a supply of air, as required.

[0055] If a fire does break out for whatever reason, the communication means also allows for the system to be able to send commands directly to safety systems provided within the environment, for example, fire suppression systems 15. Further, alert devices may also be installed as part of the present system, such as visual beacons 17, or audible sirens 19 — either or both of which may be activated when an event is detected. Depending on the type or severity of the event which is detected, the system may activate the alert devices in different variations / combinations to signify the type and / or severity of the event, such as by activated either the visual beacon 17 or the siren 19, or both simultaneously. In other embodiments, the audible alert devices 19 may output different sounding audible alerts, according to the detected event. Display means, in the form of a display screen can be provided to relay live visual data from the cameras 11 , 13, and may be provided in a number of ways. In some examples small display screens may be located on each camera head which can enable a user to see exactly what is being viewed by cameras and physically adjust the camera head such that it is directed exactly where it is required. This can be very useful as installation of such camera heads in industrial settings can provide challenges, and it becomes difficult for a user installing the camera to know exactly where the cameras are directed — the provision of a live display screen on the camera head, allowing the user to see where the camera head is being directed, alleviates this problem. Additionally, or alternatively, display means may be provided in a location remote from said camera head, for example, in a control room. In other examples, a display screen may be provided in the form of a downloadable application on a mobile device, and the communication means can be provided to communicate live visual and / or thermal data acquired from the cameras to that display. The display means may also provide thereon information pertaining to the status of the system, for example, temperature of the first and / or second camera means, air flow status, detected events etc.

[0056] Finally, the system further includes data storage means incorporated therewith, thereby enabling all thermal and visual data to be stored, replayed and reviewed at a later time, as required by a user. The system may be arranged to periodically send reports of incidents / events to users for review and analysis, enabling them to determine if there is an increased risk of such incidents or events occurring at a particular location or locations. This also enables personnel to review each event that has been detected by the system and determine whether or not they are deemed to be acceptable. This ensures the machine learning algorithm can be continuously updated and taught over time. Machine vision software is incorporated with the computing means, and which includes a neural network model. As data is captured and received by the second camera 13, that data is compared with the neural network model and any anomalies / discrepancies etc. may be identified.

Claims

CLAIMS1. A fire and / or event detection system, said system including: at least first camera means, configured to operate in a range between 700nm to 15,000nm; at least second camera means, configured to operate in a range between 380nm to 750nm; characterized in that the system includes computing means provided in communication with at least the second camera means, said computing means including a machine learning algorithm arranged to process and classify data obtained from at least the second camera means, the computing means arranged to receive and process said data and discern and detect the occurrence of an event within the field of vision of at least the second camera means.

2. A system according to claim 1 , wherein said computing means is in communication with both the first and the second camera means and data from the first camera means is processed alongside data from the second camera means, to discern and detect the occurrence of an event within the field of vision of the first and second camera means.

3. A system according to claim 1 , wherein said machine learning algorithm is programmed to recognize a standard set of working conditions and / or parameters, and thus detect events and / or occurrences falling outside of said conditions and / or parameters.

4. A system according to claim 1 , wherein said first camera means is configured to operate in a range between 700nm to l ,400nm and / or between 8,000nm to 15,000nm.

5. A system according to claim 1 , wherein the second camera means is provided as a lens-less camera means.

6. A system according to claim 1 , wherein said machine learning algorithm is programmed to divide the field of view of the first and / or second camera means into a grid comprising two or more sections and subsequently analyse each section sequentially, in use.

7. A system according to claim 2, wherein where a thermal event is detected by the first camera means, the machine learning algorithm is programmed to clarify and / or confirm said event using the second camera means, which will analyse / review the event at the specific location in its field of view, in use.

8. A system according to claim 1 , wherein the first camera means and the second camera means are provided to be located in a single camera head.

9. A system according to claim 8, wherein the system includes a plurality of camera heads located in a plurality of positions in the environment being monitored, each camera head comprising first camera means and second camera means therein.

10. A system according to claim 8, wherein the or each camera head may include further sensing means, said further sensing means including temperature sensing means and / or air flow sensing means.

11. A system according to claim 8, wherein the or each camera head includes air supply, air circulation and / or air blowing means provided therewith.

12. A system according to claim 11, wherein actuation means are provided which are arranged, in use, to activate the supply of air on to one or more lens of the first and / or second camera means, and / or circulate air through a housing of the or each camera head, said actuation means provided in communication with temperature and / or air flow sensing means, such that when a predetermined threshold is met, the actuation means activate a supply of air, in use.

13. A system according to claim 1, wherein the system further includes communication means, arranged to enable communication between at least the first and second camera means, and the computing means.

14. A system according to claim 13, wherein said communication means are provided to communicate live visual and / or thermal data acquired from the first and / or second camera means to the computing means, which is arranged to analyse and detect events and / or occurrences falling outside of a standard set of working conditions and / or parameters, in use.

15. A system according to claim 1, wherein the system further includes display means, provided to relay live visual data from the first and / or second camera means.

16. A system according to claim 15, wherein the display means are located on a camera head in which the first and second camera means are located.

17. A system according to claim 1, wherein the system further includes alert means, said alert means in communication with communication means of the system and arranged to initiate an alert, in use, when there has been detected an event and / oroccurrence falling outside of a standard set of working conditions and / or parameters.

18. A system according to claim 1 , wherein said computing means are arranged to permit images / data from the first camera means to be overlayed and / or paralleled with data from the second camera means.

19. A system according to claim 1 , wherein said computing means includes machine vision software incorporated therewith, said software includes an artificial neural network model and as data is captured and received by the second camera means, in use, said data is compared with the neural network model and any anomalies / discrepancies etc. identified.

20. A method of detecting a fire and / or event occurring using a fire and / or event detection system as defined above, said method including the steps of: providing at least first camera means, configured to operate in a range between 700nm to 15,000nm, and at least second camera means, configured to operate in a range between 380nm to 750nm, of the system to observe and monitor a location or environment; receiving and processing data from at least the second camera means and, utilising the machine learning algorithm incorporated into the computing means of the system, processing and classifying the data to discern and detect the occurrence of an event within the field of vision of the at least second camera means.

21. A method according to claim 20, wherein said computing means is in communication with both the first and the second camera means and data from the first camera means isprocessed alongside data from the second camera means to discern and detect the occurrence of an event within the field of vision of the first and second camera means.

22. A method according to claim 20, wherein the machine learning algorithm is programmed to recognize a standard set of working conditions and / or parameters, and thus detect events and / or occurrences falling outside of said conditions and / or parameters.

23. A method according to claim 20, wherein said machine learning algorithm is programmed to divide the field of view of the first and / or second camera means into a grid comprising two or more sections and subsequently analyse each section sequentially.

24. A method according to claim 20, wherein where a thermal event is detected by the first camera means, the machine learning algorithm is programmed to clarify and / or confirm said event using the second camera means, which will analyse / review the event at the specific location in its field of view.

25. A method according to claim 20, wherein communication means provided with the system, upon detection of an event and / or occurrence falling outside of a standard set of working conditions and / or parameters, will perform one or more of the following actions: communicate the detection of said event and / or occurrence to a remote location and / or user or users, via email, SMS messaging, digital signal, push notifications through a mobile application, and / or other forms of electronic communication;communicate “shut-down”, “slow-down”, “reject product” or other such commands to machinery or a section of machinery for which there has been detected an event and / or occurrence; communicate to a safety system installed in the location or environment to activate, or activate a certain section, as required; and / or communicate with alert means provided associated with the system to initiate an audio and / or visual alert when there has been detected an event and / or occurrence.

Images