System and method for detecting tanks in a railway environment

Abstract

In one embodiment, the method comprises the steps of identifying a tank within an image of a railway environment by an image detection tool and identifying a railroad track within an image of a railway environment by an image detection tool. The method also comprises the steps of determining a distance between the tank and the railroad track by an image detection tool and comparing the distance between the tank and the railroad track by an image detection tool with a predetermined threshold distance. The method further comprises the step of determining, by an image detection tool, whether the tank poses a risk to the railway environment in response to comparing the distance between the tank and the railroad track with the predetermined threshold distance.

Description

Technology Field

[0001] The present disclosure generally relates to detecting tanks, and more specifically to a system and method for detecting tanks in a railway environment. Background Technology

[0002] Certain railway companies prohibit the storage of hazardous materials, such as anhydrous ammonia, on their properties. Railroad inspectors can identify storage tanks containing hazardous materials through routine property inspections. However, due to limited resources, such as time and personnel, tanks storing hazardous materials on railway property may go undetected. means of solving the problem

[0003] According to one embodiment, the method comprises the steps of identifying a tank within an image of a railway environment by an image detection tool and identifying a railroad track within an image of a railway environment by an image detection tool. The method also comprises the steps of determining a distance between the tank and the railroad track by an image detection tool and comparing the distance between the tank and the railroad track by an image detection tool with a predetermined threshold distance. In response to comparing the distance between the tank and the railroad track with the predetermined threshold distance, the method further comprises the step of determining, by an image detection tool, whether the tank poses a risk to the railway environment.

[0004] In certain embodiments, the method includes the step of determining that the distance between the tank and the railway track is less than or equal to a predetermined threshold distance. For example, the predetermined threshold distance may be 50 feet. In some embodiments, the method includes the step of classifying the risk as high risk, medium risk, or low risk based on the distance between the tank and the railway track by an image detection tool. For example, high risk may be associated with a distance between 0 and 20 feet, medium risk may be associated with a distance between 21 and 30 feet, and low risk may be associated with a distance between 31 and 50 feet.

[0005] In certain embodiments, the method includes the step of generating an alert by an image detection tool. The alert may include at least one of: a hazard, an indication of the hazard level associated with the hazard, a description of the tank, the location of the tank, the distance between the tank and the railway track, an image of the railway environment, the time the image of the railway environment was captured, and / or an indication of the date the image of the railway environment was captured. In some embodiments, the method includes the step of training the image detection tool to identify the tank using a machine learning model.

[0006] In certain embodiments, images are captured by a camera attached to the train car while the train car moves along a railway track through a railway environment. The train car may be a locomotive or a geometry car. In some embodiments, the images include a railway track and one or more other railway tracks, the railway track is closer to the tank than one or more other railway tracks, and the distance between the tank and the railway track is the shortest distance measured when viewed in a plane from the centerline of the outer rail of the railway track to the outer edge of the tank.

[0007] According to another embodiment, the device comprises one or more processors and a memory for storing instructions, and when executed by one or more processors, the instructions cause one or more processors to perform operations including identifying a tank within an image of a railway environment and identifying a railway track within an image of a railway environment. The operation also includes the steps of determining the distance between the tank and the railway track and comparing the distance between the tank and the railway track with a predetermined threshold distance. The operation further includes the step of determining whether the tank poses a risk to the railway environment in response to comparing the distance between the tank and the railway track with the predetermined threshold distance.

[0008] According to another embodiment, one or more computer-readable storage media contain instructions, and when executed by a processor, the instructions cause the processor to perform an operation including the steps of identifying a tank within an image of a railway environment and identifying a railway track within an image of a railway environment. The operation also includes the steps of determining the distance between the tank and the railway track and comparing the distance between the tank and the railway track with a predetermined threshold distance. The operation further includes the step of determining whether the tank poses a risk to the railway environment in response to comparing the distance between the tank and the railway track with the predetermined threshold distance.

[0009] The technical benefits of a specific embodiment of the present disclosure may include one or more of the following. The specific system and method described herein include an image recognition tool that automatically detects tanks located around railway tracks, which increases efficiency by reducing the manpower and time required for manual inspection. Safety is enhanced by automatically detecting tanks located around railway tracks to identify potential hazardous materials around the tracks. In a specific embodiment, the automatic detection of tanks reduces environmental risks caused by hazardous materials.

[0010] Other technical advantages will be readily apparent to those skilled in the art from the following drawings, description, and claims. Furthermore, while specific advantages have been listed above, various embodiments may include all or part of the listed advantages, or none at all. Brief explanation of the drawing

[0011] To aid in understanding the content of the present disclosure, refer to the following description taken in connection with the attached drawings. FIG. 1 illustrates an exemplary system for detecting a tank in a railway environment; FIG. 2 illustrates an exemplary image that can be used by the system of FIG. 1; FIG. 3 illustrates an exemplary method for detecting a tank in a railway environment; and FIG. 4 illustrates an exemplary computer system that can be used by the system and method described in this specification. Specific details for implementing the invention

[0012] FIGS. 1 through 4 illustrate exemplary systems and methods for detecting tanks in a railway environment. FIG. 1 illustrates an exemplary system for detecting tanks in a railway environment, and FIG. 2 illustrates an exemplary image that can be used by the system of FIG. 1. FIG. 3 illustrates an exemplary method for detecting tanks in a railway environment. FIG. 4 illustrates an exemplary computer system that can be used by the systems and methods described herein.

[0013] FIG. 1 illustrates an exemplary system (100) for detecting tanks in a railway environment. The system (100) of FIG. 1 includes a network (110), a railway environment (120), a railroad track (130) (i.e., railroad track (130a) and railroad track (130b)), a train car (140), a camera (150), an image detection tool (180), and a training module (190). The system (100) or part thereof may be associated with an entity, which may include any entity such as a company, a firm (e.g., a railway company, a transport company, etc.) or a government agency (e.g., a transport department, a public safety department, etc.) capable of detecting tanks in a railway environment. Although the exemplary embodiment of FIG. 1 is associated with a railroad system, the system (100) may be associated with any suitable transportation system (e.g., a vehicle / road, a ship / waterway, etc.). For example, the system (100) may be used to detect tanks in very close proximity (e.g., within 50 feet) to a road, a waterway, etc. Elements of the system (100) may be implemented using any suitable combination of hardware, firmware, and software. For example, one or more components of the system (100) may use one or more components of FIG. 4.

[0014] The network (110) may be any type of network that facilitates communication between components of the system (100). For example, the network (110) may connect the camera (150) of the system (100) to the image detection tool (180) of the system (100). As another example, the network (110) may connect the image detection tool (180) of the system (100) to the training module (190) of the system (100). One or more parts of the network (110) may include an ad-hoc network, an intranet, an extranet, a virtual private network (VPN), a local area network (LAN), a wireless LAN (WLAN), a wide area network (WAN), a wireless WAN (WWAN), a Metropolitan Area Network (MAN), part of the Internet, part of the Public Switched Telephone Network (PSTN), a cellular telephone network, a 3G network, a 4G network, a 5G network, a Long Term Evolution (LTE) cellular network, a combination of two or more of these, or other suitable types of networks. One or more parts of the network (110) may include one or more access (e.g., mobile access), core and / or edge networks. The network (110) may be any communication network such as a private network, a public network, a connection via the Internet, a mobile network, a Wi-Fi network, a Bluetooth network, etc. The network (110) may include cloud computing capabilities. One or more components of the system (100) may communicate through the network (110).For example, the camera (150) may communicate through the network (110), including transmitting information (e.g., an image (182)) to the image detection tool (180) and / or receiving information (e.g., an alarm) from the image detection tool (180). As another example, the training module (190) may communicate through the network (110), including transmitting information (e.g., a training image (192)) to the image detection tool (180).

[0015] The railway environment (120) of the system (100) is an area containing one or more railroad tracks (130). The railway environment (120) may be associated with a division, subdivision, district, region, branch, etc. A division is a railway section under the supervision of a superintendent. A subdivision is a smaller part of a division. A subdivision may be a crew area and / or a branch line. In the embodiment illustrated in FIG. 1, the railway environment (120) includes a railroad track (130), a train car (140), a camera (150), and a tank (170) (i.e., tank (170a) and tank (170b)).

[0016] The railway track (130) of the system (100) is a structure that enables the train vehicle (140) to move by providing a surface on which the wheels of the train vehicle (140) roll. In certain embodiments, the railway track (130) includes rails, fasteners, railroad ties, ballast, etc. The train vehicle (140) is a vehicle that transports cargo and / or passengers in a railway transport system. In certain embodiments, the train vehicle (140) is connected to other train vehicles to form a train. The train vehicle (140) may be a geometry car (e.g., a track geometry car), a locomotive, a passenger car, a cargo car, a box car, a flat car, a tank car, etc.

[0017] In the exemplary embodiment of FIG. 1, the train car (140) is moving along the railway track (130b) in a direction of travel (160) (e.g., south). In some embodiments, the railway track (130a) of the railway environment (120) is adjacent (e.g., parallel) to the railway track (130b) of the railway environment (120). In certain embodiments, the train car (140) may move along the railway track (130a) in a direction opposite to the direction of travel (160) (e.g., north). In some embodiments, the train car (140) may move along the railway track (130a) in a direction of travel (160) (or in a direction opposite to the direction of travel (160).

[0018] The camera (150) of the system (100) is a component that automatically captures a still or moving image (182). The camera (150) may include one or more lenses, sensors, optical devices, lighting elements, etc. In a specific embodiment, the camera (150) provides a high resolution, 360-degree visual range. The camera (150) may capture the image (182) in real-time or near real-time. For example, the camera (150) may capture an image (182) of an object (e.g., a tank (170a) and / or a tank (170b)) in the railway environment (120) and communicate the image (182) to a component outside the camera (150) (e.g., an image detection tool (180) or a training module (190)) for a shorter time than a predetermined time (e.g., 1 second, 5 seconds, 10 seconds, or any other suitable time). In some embodiments, the camera (150) tracks the time and / or date when each image (182) was captured. For example, the camera (150) may record the time and / or date when each captured image (182) was captured on each captured image (182).

[0019] In a specific embodiment, the camera (150) automatically captures an image (182) of the railway environment (120) of the system (100). The camera (150) can automatically capture a stationary or moving image (182) while the train car (140) moves along the railway track (130b). The camera (150) can automatically capture any appropriate number of stationary or moving images (182). For example, the camera (150) can automatically capture a predetermined number of images (182), such as per second, per minute, or per hour. In a specific embodiment, the camera (150) captures a sufficient number of images (182) to capture the entire length of the railway track (130b) within a predetermined area (e.g., a sub-point, a sub-point, or other appropriate area).

[0020] A camera (150) of the system (100) is attached to a train car (140). The camera (150) may be attached to the train car (140) at any suitable location that provides a clear view of the railway track (130b). For example, the camera (150) may be attached to the rear end of the train car (140) (e.g., rear windshield, roof, etc.) to provide a view toward the rear of the railway track (130b). As another example, the camera (150) may be attached to the front end of the train car (140) (e.g., front windshield) to provide a view toward the front of the railway track (130b).

[0021] The tanks (170) of the system (100) (i.e., tank (170a) and tank (170b)) are storage containers that hold materials (e.g., liquids, compressed gases, chemicals, etc.) for short-term or long-term storage. In a specific embodiment, the tanks (170) store hazardous materials.For example, the tank (170) is toxic and / or poison inhalation hazard (e.g., anhydrous ammonia), allyl chloride, benzene, carbon disulfide, carbon tetrachloride, chlorobenzene, chlorobenzyl chloride, chloroform, dichloropropene, epichlorohydrin, ethyl chloride, ethylene dichloride, methyl chloroform (trichloroethane), methylene chloride (dichloromethane), methylene chloride / chloroform mixture, o-dichlorobenzene, Perchloroethylene (tetrachloroethylene), perchloroethylene / trichloroethylene mixture, trichloroethylene, flammable gases, non-flammable gases, poison gases, fuel, oil, propane, combinations thereof, or other hazardous substances may be stored. In a specific embodiment, the tank (170) is an above-ground storage tank.

[0022] The image detection tool (180) of the system (100) is a component that inspects, evaluates, and / or processes a stationary or moving image (182). The image detection tool (180) can inspect the image (182) for objects. Objects may include one or more railway tracks (130), tanks (170), power plants, power lines, road crossings, road crossing stop bars, milestone signs, speed signs, railway signals, etc. In some embodiments, the image detection tool (180) may use an image detection algorithm to analyze objects within the image (182). The image detection algorithm may recognize objects in the image (182) and classify the objects using image processing techniques and / or pattern recognition techniques. In certain embodiments, the image detection tool (180) uses a cloud computing platform (e.g., Microsoft Azure, Amazon Web Services (AWS), etc.) to inspect, evaluate, and / or process the image (182).

[0023] In a specific embodiment, the image detection tool (180) uses an image detection algorithm to analyze objects (e.g., tanks (170), power plants, power lines, road crossings, road crossing stop bars, milestone signs, speed signs, railway signals, etc.) in an image (182) for specific characteristics (e.g., size, shape, location, etc.). For example, the image detection tool (180) may determine the length of each tank (170), the width of each tank (170), the height of each tank (170) relative to the ground, the shape of each tank (170) (e.g., cylindrical), the roof type of each tank (170) (e.g., fixed flange type or floating roof), etc. As another example, the image detection tool (180) may determine the position of the crossing stop bar (e.g., up or down). In some embodiments, the image detection tool (180) may classify each object based on the characteristics of each object. For example, the image detection tool (180) can classify each tank (170) into one of the following categories: anhydrous ammonia tank, methane tank, water tank, etc.

[0024] In some embodiments, the image detection tool (180) may determine whether each tank (170) contains hazardous substances based on the characteristics and / or classification of each tank (170). For example, the image detection tool (180) may determine that tank (170a) contains hazardous substances, such as anhydrous ammonia, based on the size and shape of tank (170a). As another example, the image detection tool (180) may determine that tank (170b) contains non-hazardous substances, such as water, based on the size and shape of tank (170b). In certain embodiments, the image detection tool (180) may determine whether each tank (170) poses a risk to the railway environment (120) based at least partially on the characteristics and / or classification of each tank (170).

[0025] In some embodiments, the image detection tool (180) uses an image (182) to determine the location of one or more objects. The location may represent a physical (e.g., geographical) location within the railway environment (120). For example, the image detection tool (180) may use the image (182) to determine the geographical location (e.g., Global Positioning System (GPS) coordinates) of an object (e.g., railway track (130a), railway track (130b), tank (170a, 170b), etc.). The location may represent any point (e.g., center or edge) within an area in a plan view of the object.

[0026] In some embodiments, the image detection tool (180) determines the location of an object in the image (182) relative to other objects in the image (182). For example, the image detection tool (180) may determine the distance between the location of a first object in the image (182) (e.g., a railway track (130b)) and the location of a second object in the image (182) (e.g., a tank (170a)). The distance between the two objects may represent a physical (e.g., geographical) distance in the railway environment (120). For example, the image detection tool (180) may use the image (182) to determine that the distance between the tank (170a) and the railway track (130b) in the railway environment (120) is 20 feet. In certain embodiments, the distance is measured in a plan view from the edge of the tank (170a) closest to the railway track (130b) to the centerline of the railway track (130b). In some embodiments, the distance is measured in a plan view from the edge of the tank (170a) closest to the railway track (130b) to the centerline of the railway track (130b). The distance between the railway track (130b) and each tank (170) may represent the shortest distance in a plan view between the centerline of the railway track (130b) (or the centerline of the outer rail of the railway track (130b)) and each tank (170).

[0027] In some embodiments, the image detection tool (180) compares the distance between each tank (170) in the railway environment (120) and the railway track (130b) with a predetermined threshold distance to determine whether each tank (170) represents a danger to the railway environment (120). If the tank (170a) is located within the predetermined threshold distance to the railway track (130b), the tank (170a) represents a danger to the railway environment (120). In certain embodiments, the predetermined threshold distance is 50 feet. For example, the image detection tool (180) may determine that the geographical distance between the centerline of the railway track (130b) and the tank (170a) is less than or equal to the predetermined threshold distance of 50 feet, and may determine based on the comparison that the tank (170) represents a danger to the railway environment (120).

[0028] In certain embodiments, each risk is classified according to a risk level. A risk level is the likelihood that the risk will actually cause harm to the railway environment (120). Risk levels may include a high risk level, a medium risk level, a low risk level, or other representative levels identifying potential risks. In some embodiments, different predetermined threshold distances are associated with different risk levels. For example, a threshold distance between 0 and 20 feet may be associated with a high risk level, a threshold distance between 21 and 30 feet may be associated with a medium risk level, and a threshold distance between 31 feet and 50 feet may be associated with a low risk level. As another example, a threshold distance between 0 and 25 feet may be associated with a high risk level, a threshold distance between 26 and 35 feet may be associated with a medium risk level, and a threshold distance between 36 feet and 55 feet may be associated with a low risk level. As another example, a threshold distance between 0 and 15 feet may be associated with a high risk level, a threshold distance between 16 and 25 feet may be associated with a medium risk level, and a threshold distance between 26 and 40 feet may be associated with a low risk level. The image detection tool (180) may generate one or more alarms (184) in response to a determination that the tank (170a) and / or tank (170b) indicates a risk / danger to the railway environment (120).

[0029] The image detection tool (180) may transmit one or more alarms (184) to one or more components of the system (100). Each alarm (184) indicates a risk notification regarding the railway environment (120). The alarms (184) may be transmitted in any suitable format (e.g., email, voicemail, fax, text message, post on a website, etc.). In a specific embodiment, the image detection tool (180) transmits one or more alarms (184) to an operator (e.g., manager, engineer, supervisor, etc.). For example, the image detection tool (180) may determine that the tank (170a) indicates a high risk to the railway environment (120) and transmit the high risk alarm (184) to one or more operators and / or components of the system (100). As another example, the image detection tool (180) may determine that the tank (170b) indicates a moderate risk to the railway environment (120) and may transmit an alarm (184) indicating a moderate risk to one or more operators and / or components of the system (100). In a specific embodiment, the alarm (184) may be transmitted to a network operations center.

[0030] In a specific embodiment, an alarm (184) generated by an image detection tool (180) may include: a description of one or more tanks (170) located within a danger zone (e.g., within 50 feet) of the railway track (130b), the distance between the railway track (130b) and one or more tanks (170), an image (184) of the tank (170), the location of the tank (170) (e.g., the GPS location of the tank (170)), the time the tank (170) was captured by the camera (150) of the train car (140), the date the tank (170) was captured by the camera (150) of the train car (140), the identification of the train car (140), an indication of the direction of movement (160) of the train car (140).

[0031] In some embodiments, information generated by the system (100) may be combined with information generated by one or more other systems (e.g., a real estate system, a geographic information system (GIS), etc.). For example, information generated by the image detection tool (180) of the system (100) (e.g., tank identification, tank location, etc.) may be combined with information generated by the real estate system (e.g., land ownership, contracts, etc.) and / or information generated by the GIS (e.g., digitized map views). In certain embodiments, one or more components of the system (100) may generate a report containing combined and / or compared information. The report may provide a comprehensive view of the railway environment (120).

[0032] The training module (190) of the system (100) is an application that trains one or more models to detect objects (e.g., a tank (170), a power plant, a power line, a road crossing, a road crossing stop bar, a milestone sign, a speed sign, a railway signal, etc.). For example, the training module (190) can train one or more models to detect objects of interest (e.g., a tank (170), a power plant, a power line, a road crossing, a road crossing stop bar, a milestone sign, a speed sign, a railway signal, etc.) in an image (182) (e.g., a digital image, a video, etc.). The training module (190) can use a training image (192) to train one or more models to detect objects of interest. The training image (192) is an image containing objects of interest. For example, the training image (192) may include an image of a tank (170) in a railway environment (120). In a specific embodiment, the object of interest (e.g., tank (170)) is superimposed on a training image (192). For example, an existing image of the tank (170) can be pasted onto an image (182) containing a railway track (130). In a specific embodiment, the training module (190) receives the training image (192) from the administrator of the system (100). The quantity and variety of training images (192) utilized by the training module (190) depend on the desired accuracy, the number of classifications of the tank (170), the availability of training images (192) for each classification, etc.

[0033] In a specific embodiment, the training module (190) uses one or more machine learning approaches to train the model. A machine learning approach uses an algorithm to analyze data, learn from the data, and make informed decisions based on the learned information. A machine learning approach may include one or more of: the Viola-Jones object detection framework, Scale-invariant feature transform (SIFT), and histogram of oriented gradients (HOG) functions. A specific machine learning approach includes a deep learning approach. A deep learning approach can learn unsupervised data that is unstructured or unlabeled. Deep learning approaches may include one or more of: You Only Look Once (YOLO), local proposals (e.g., R-CNN, Fast R-CNN, Faster R-CNN, etc.), Single Shot MultiBox Detector (SSD), Single-Shot Refinement Neural Network for Object Detection (RefineDet), Retina-Net, deformable convolutional network, etc.

[0034] When in operation, the training module (190) of the system (100) trains an image detection tool (180) to identify a tank (170) in an image (182). A camera (150) is attached to a train car (140) (e.g., a locomotive or a geometry car) and captures an image (182) of the railway environment (120) while the train car (140) moves along a railway track (130b) in a direction of travel (160). The camera (150) transmits the image (182) to the image detection tool (180). The image detection tool (180) scans the image (182) and identifies a tank (170a) and a tank (170b) within the image (182). An image detection tool (180) identifies a railway track (130b) within an image (182) and determines a first distance (e.g., 18 feet) between a tank (170a) and a railway track (130b) and a second distance (e.g., 30 feet) between a tank (170b) and a railway track (130b). The image detection tool (180) compares the first and second distances with a predetermined threshold distance (e.g., 50 feet) and determines that the first and second distances are smaller than the predetermined threshold distance. Based on this determination, the image detection tool (180) determines that the tank (170a) and the tank (170b) represent a risk to the railway environment (120). The image detection tool classifies the risk as low risk, medium risk, or high risk and generates an alarm (184) indicating the risk and the associated risk level. In this way, the system (100) can be used to alert one or more individuals to the danger of the railway environment (120) because the tank (170) is located in close proximity to the railway track (e.g., within 50 feet), which can initiate corrective action to reduce or eliminate the danger.

[0035] FIG. 1 illustrates a specific arrangement of a network (110), a railway environment (120), a railway track (130), a train car (140), a camera (150), a tank (170), an image detection tool (180), and a training module (190), but the present disclosure considers any suitable arrangement of the network (110), the railway environment (120), the railway track (130), the train car (140), the camera (150), the tank (170), the image detection tool (180), and the training module (190). For example, the tank (170b) may be positioned closer to the railway track (130b) than the tank (170a). As another example, the train car (140) may move along the railway track (130a) instead of the railway track (130b). As yet another example, the train car (140) may be moving in the opposite direction to the direction of movement (160). As another example, the image detection tool (180) and the training module (190) can be combined into a single component.

[0036] FIG. 1 illustrates a specific number of networks (110), a railway environment (120), a railway track (130), a train car (140), a camera (150), a tank (170), an image detection tool (180), and a training module (190), but the present disclosure considers any suitable number of networks (110), a railway environment (120), a railway track (130), a train car (140), a camera (150), a tank (170), an image detection tool (180), and a training module (190). For example, the system (100) may include more or fewer than two railway tracks (130). As another example, the system (100) may include more or fewer than two tanks (170).

[0037] FIG. 2 illustrates an exemplary image (182) that can be used by the system (100) of FIG. 1. The image (182) shows an overview of the railway environment (120) of FIG. 1 at a specific moment. The image (182) includes a railway track (130) (i.e., railway track (130a) and railway track (130b)), a tank (170a), and a tank (170b). In the exemplary embodiment of FIG. 2, the railway track (130a) is adjacent (e.g., parallel) to the railway track (130b). In a specific embodiment, the camera (150) of FIG. 1 automatically captures the image (182) as the train car (140) moves along the railway track (130b) through the railway environment (120).

[0038] The image (182) of FIG. 2 includes a number of labels. In some embodiments, the image detection tool (180) of FIG. 1 generates one or more labels for the image (182). The labels represent information associated with the image (182). For example, the image detection tool (180) may generate one or more labels for the image (182) that identify one or more objects (e.g., railway track (130a), railway track (130b), tank (170a), tank (170b), etc.). As another example, the image detection tool (180) may generate one or more labels for the image (182) that identify one or more risks associated with the railway environment depicted in the image (182). In certain embodiments, the labels identify a risk level associated with the risk (e.g., high risk, medium risk, low risk, etc.). As another example, the image detection tool (180) may generate one or more labels for the image (182) that provide additional information about the image (182). The additional information may include a description of an object (e.g., tank (170a), tank (170b), etc.), the location of the object (e.g., GPS location, mile marker near the object, etc.), the distance between objects, the time the image (182) of the railway environment was captured, the date the image (182) of the railway environment was captured, etc. In some embodiments, one or more labels are superimposed on the image (182). In the exemplary embodiment of FIG. 2, the labels include tank identification (210), track identification (220), distance (230), and alarm (184).

[0039] Tank identification (210) of image (182) (i.e., tank identification (210a) and tank identification (210b)) is a label used to identify a tank within image (182). Tank identification (210) is represented by a dotted box in the exemplary embodiment of FIG. 2, but tank identification (210) can be represented in any suitable manner. For example, tank identification (210) can be represented by a solid box, a highlighted area, letters, numbers, a combination thereof, etc. In a particular embodiment, the image detection tool (180) of FIG. 1 identifies a tank (170a) within image (182) and labels the tank (170a) with tank identification (210a). Similarly, the image detection tool (180) can identify a tank (170b) within image (182) and label the tank (170b) with tank identification (210b). The tank identification (210a) in image (182) represents the outer perimeter in the plan view of the tank (170a). The tank identification (210b) in image (182) represents the outer perimeter in the plan view of the tank (170b).

[0040] Track identification (220) of image (182) is a label used to identify railway tracks within image (182). Track identification (220) is represented by a dotted line in the exemplary embodiment of FIG. 2, but track identification (220) can be represented in any suitable manner. For example, track identification (220) can be represented by a solid line, a bold line, a highlighted area, a letter, a number, a combination thereof, etc. In a specific embodiment, the image detection tool (180) of FIG. 1 identifies railway track (130a) and railway track (130b) within image (182) and labels railway track (130a) and railway track (130b) with track identification (220). Each track identification (220) of image (182) represents the centerline of a rail. For example, two dotted lines along the railway track (130a) of image (182) represent the centerlines of the inner and outer rails for the railway track (130a). As another example, the two dashed lines along the railway track (130b) in image (182) represent the centerlines of the inner and outer rails for the railway track (130b).

[0041] The distance (230) in the image (182) is a label used to indicate the distance between the tank (170) and the railway track (130). The distance (230) is represented as a solid line with an arrow at each end in the exemplary embodiment of FIG. 2, but the distance (230) may be represented in any suitable manner. For example, the distance (230) may be indicated as a dashed line with an arrow at each end, a solid line with a scale at each end, a thick line, letters, numbers, a combination thereof, etc. In a particular embodiment, the distance (230a) between the tank (170a) and the railway track (130b) is the shortest distance measured from the centerline of the outer rail of the railway track (130b) to the outer edge of the tank (170a) in the plan view. In some embodiments, the distance (230b) between the tank (170b) and the railway track (130b) is the shortest distance measured from the centerline of the outer rail of the railway track (130b) to the outer edge of the tank (170b) in a plan view. The distance (230) in image (182) represents the actual (e.g., geographical) distance between the tank (170) and the railway track. The distance (230) may be expressed using metric units (e.g., meters, centimeters, etc.), standard units (e.g., feet, inches, etc.), etc. In certain embodiments, the distance (230) may be approximate to the actual field measurement distance (e.g., within 3 percent, 5 percent, or 10 percent).

[0042] The alarm (184) of the image (182) is a label used to identify danger within the image (182). In the exemplary embodiment of FIG. 2, the alarm (184) includes a high danger alarm (184a) associated with the tank (170a) and a medium danger alarm (184b) associated with the tank (170b). In a specific embodiment, the image detection tool (180) of FIG. 1 generates the alarm (184) in response to comparing a distance (230) with one or more predetermined threshold distances. For example, the image detection tool (180) may compare the distance (230a) (e.g., 18 feet) between the tank (170a) and the railway track (130b) with a first predetermined threshold distance (e.g., 50 feet) to determine that the tank (170a) represents a danger to the railway environment. Similarly, the image detection tool (180) can compare the distance (230b) (e.g., 30 feet) between the tank (170b) and the railway track (130b) with a first predetermined threshold distance (e.g., 50 feet) to determine that the tank (170b) indicates a risk to the railway environment. As another example, the image detection tool (180) can compare the distance (230a) (e.g., 18 feet) between the tank (170a) and the railway track (130b) with a second predetermined threshold distance (e.g., 20 feet) to determine that the tank (170a) indicates a high risk to the railway environment. Similarly, the image detection tool (180) can compare the distance (230b) (e.g., 30 feet) between the tank (170a) and the railway track (130b) with a second predetermined threshold distance (e.g., 20 feet) to determine that the tank (170a) indicates a moderate risk to the railway environment.

[0043] In response to this decision, the image detection tool (180) may generate an alarm (184) to notify one or more individuals (e.g., managers, engineers, etc.) of the risk and / or risk level. The alarm (184) is represented by a combination of letters, numbers, and symbols in the exemplary embodiment of FIG. 2, but the alarm (184) may be represented in any suitable manner. For example, the alarm (184a) may identify the tank (170a) as high risk by placing a red box around the tank (170a). As another example, the alarm (184b) may identify the tank (170b) as medium risk by placing an orange box around the tank (170b).

[0044] FIG. 2 illustrates a specific arrangement of a railway track (130), a tank (170), a tank identification (210), a track identification (220), a distance (230), and an alarm (184) within an image (182), but the present disclosure considers any suitable arrangement of a railway track (130), a tank (170), a tank identification (210), a track identification (220), a distance (230), and an alarm (184) within an image (182). For example, a tank (170a) and / or a tank (170b) may be located closer to the railway track (130a) than to the railway track (130b). As another example, the track identification (220) may represent the centerline of the railway track (130a) rather than the centerline of the rail of the railway track (130a). As another example, the alarm (184) may be located near the bottom of the image (184) in the plan view rather than near the top of the image (184).

[0045] FIG. 2 illustrates a specific number of railway tracks (130), tanks (170), tank identification (210), track identification (220), distances (230), and alarms (184) within an image (182), but the present disclosure considers any suitable number of railway tracks (130), tanks (170), tank identification (210), track identification (220), distances (230), and alarms (184) within the image (182). For example, the image (182) may contain more than two or fewer tanks (170). As another example, the image (182) may contain more than two or fewer alarms (184). The image (182) of FIG. 2 is associated with a railway system, but the image (182) may be associated with any suitable transport system (e.g., vehicles / roads, ships / waterways, etc.).

[0046] FIG. 3 illustrates an exemplary method (300) for detecting a tank in a railway environment. The method (300) begins at step (305). In step (310), a training module (e.g., the training module (190) of FIG. 1) trains an image detection tool (e.g., the image detection tool (180) of FIG. 1) to identify a tank (e.g., a tank (170) in the railway environment (120) of FIG. 1) in a railway environment. In a specific embodiment, the training module uses a training image (e.g., the training image (192) of FIG. 1) to train the image detection tool to detect a tank. The training module may use a deep learning approach and / or a machine learning approach to train the image detection tool to detect a tank. Then, the method (300) moves from step (310) to step (315).

[0047] In step (315) of method (300), an image detection tool scans an image of the railway environment (e.g., image (182) of FIG. 2). The image may be captured by a camera (e.g., camera (150) of FIG. 1) attached to a train car (e.g., train car (140) of FIG. 1) while the train car is moving along a railway track (e.g., railway track (130b) of FIG. 1) in the railway environment. In a specific embodiment, the image detection tool receives the image from the camera. Then, method (300) proceeds from step (315) to step (320).

[0048] In step (320) of method (300), an image detection tool determines whether the image contains one or more tanks. If the image detection tool does not identify a tank in the image, method (300) proceeds from step (320) to step (370), where method (300) ends. If the image detection tool identifies a tank in the image in step (320), method (300) moves from step (320) to step (325), where the image detection tool determines the location of the tank. The location of the tank may be the location of the tank in the image, a physical location (e.g., GPS location), the location of the tank relative to one or more other objects in the image, etc. In a particular embodiment, the determined location of the tank is approximate (e.g., within 10 percent) of the actual physical location of the tank. Then, method (300) moves from step (325) to step (330).

[0049] In step (330) of method (300), an image detection tool determines the location of the railway track within the image closest to the tank. For example, the image detection tool may determine the location of the centerline of the outer rail of the railway track closest to the tank (e.g., the railway track (130b) in FIG. 2). As another example, the image detection tool may determine the location of the centerline of the railway track closest to the tank. In a specific embodiment, the determined location of the railway track is approximate (e.g., within 10 percent) to the actual physical location of the railway track. Then, method (300) moves from step (330) to step (335).

[0050] In step (335) of method (300), the image detection tool determines the distance between the tank and the railway track. For example, the image detection tool can determine the physical distance (e.g., 18 feet) between the outer perimeter of the tank in a plan view and the centerline of the outer rail of the nearest railway track. As another example, the image detection tool can determine the physical distance (e.g., 20 feet) between the outer perimeter of the tank in a plan view and the centerline of the nearest railway track. Then, method (300) moves from step (335) to step (350).

[0051] In step (340) of method (300), the image detection tool determines whether the distance between the tank and the railway track is less than or equal to a first predetermined distance (e.g., 50 feet). If the image detection tool determines that the distance between the tank and the railway track is greater than the first predetermined distance, method (300) proceeds from step (340) to step (370) and method (300) ends. In step (340), if the image detection tool determines that the distance between the tank and the railway track is less than or equal to the first predetermined distance, method (300) moves from step (340) to step (345), where the image detection tool determines that the tank poses a risk to the railway environment. Then method (300) moves from step (345) to step (350).

[0052] In step (350) of method (300), the image detection tool determines whether the distance between the tank and the railway track is less than or equal to a second predetermined distance (e.g., 20 feet). If the image detection tool determines that the distance between the tank and the railway track is greater than the second predetermined distance, method (300) proceeds from step (350) to step (360). In step (350), if the image detection tool determines that the distance between the tank and the railway track is less than or equal to the second predetermined distance, method (300) moves from step (350) to step (355), where the image detection tool determines that the tank poses a high risk to the railway environment. Then method (300) moves from step (335) to step (360).

[0053] In step (360) of method (300), the image detection tool generates an alarm indicating danger (e.g., alarm (184a) or alarm (184b) of FIG. 2). For example, the image detection tool may generate an alarm indicating that the tank poses a danger (e.g., low danger or medium danger) to the railway environment if the distance between the tank and the railway track is less than or equal to a predetermined first threshold (e.g., 50 feet) but greater than a second predetermined threshold (e.g., 20 feet). As another example, the image detection tool may generate an alarm indicating that the tank poses a high danger to the railway environment if the distance between the tank and the railway track is less than or equal to a predetermined second threshold (e.g., 20 feet). Then, method (300) moves from step (360) to step (365), where the image detection tool transmits the alarm to an operator (e.g., manager, engineer, etc.). Then, method (300) moves from step (355) to step (360), where method (300) ends.

[0054] Modifications, additions, or omissions may be made to the method (300) illustrated in FIG. 3. The method (300) may include more, fewer, or different steps. For example, the method (300) may include an additional step for determining whether the tank represents a low or medium risk to the railway environment. As another example, the method (300) may include one or more additional steps for identifying the second tank within an image and determining whether the second tank represents a risk to the railway environment. As yet another example, the method (300) may include one or more steps for determining the characteristics of the tank (e.g., size, shape, etc.) and determining whether the tank contains hazardous substances based on the characteristics. One or more steps of the method (400) may be performed in real time.

[0055] The method (300) may be associated with any suitable transport system (e.g., vehicle / road, ship / waterway, etc.). The steps may be performed in parallel or in any suitable order. While discussed as specific components that complete the steps of the method (300), any suitable component may perform any step of the method (300). For example, one or more steps of the method (300) may be automated using one or more components of the computer system of FIG. 4. In a specific embodiment, one or more steps of the method (300) are determined using a machine learning approach (e.g., YOLO).

[0056] FIG. 4 illustrates an exemplary computer system that can be used by the system and method described herein. For example, the network (110), camera (150), image detection tool (180), and / or training module (190) of FIG. 1 may include one or more interface(s) (410), processing circuits (420), memory(s) (430), and / or other suitable elements. The interface (410) receives input, transmits output, processes input and / or output, and / or performs other suitable operations. The interface (410) may include hardware and / or software.

[0057] Processing circuitry (420) performs or manages the operation of the component. Processing circuitry (420) may include hardware and / or software. Examples of processing circuitry include one or more computers, one or more microprocessors, one or more applications, etc. In a specific embodiment, processing circuitry (420) executes logic (e.g., instructions) to perform an operation (e.g., an operation) such as generating an output from an input. The logic executed by processing circuitry (420) may be encoded into one or more types of non-transient computer-readable media (e.g., memory (430)). For example, the logic may include computer programs, software, computer-executable instructions, and / or instructions that can be executed by a computer. In a specific embodiment, the operation of the embodiment may be performed by one or more computer-readable media having a computer program, which stores, implements, and / or encodes the computer program and / or has a stored and / or encoded computer program.

[0058] Memory (430) (or memory unit) stores information. Memory (430) (e.g., memory (124) of FIG. 1) may include one or more non-transient, tangible, computer-readable and / or computer-executable storage media. Examples of memory (430) include computer memory (e.g., RAM or ROM), mass storage media (e.g., hard disk), removable storage media (e.g., compact disc (CD) or digital video disc (DVD)), database and / or network storage (e.g., server) and / or other computer-readable media.

[0059] Herein, a computer-readable non-transient storage medium or medium may include one or more semiconductor-based or other integrated circuits (ICs) (e.g., Field Programmable Gate Array (FPGA) or Application-Specific IC (ASIC)), a hard disk drive (HDD), a hybrid hard drive (HHD), an optical disc, an optical disc drive (ODD), a magneto-optical disc, a magneto-optical drive, a floppy diskette, a floppy disk drive (FDD), a magnetic tape, a solid-state drive (SSD), a RAM drive, a secure digital card or drive, other suitable computer-readable non-transient storage media, or, where appropriate, a suitable combination of two or more of these. A computer-readable non-transient storage medium may, where appropriate, be volatile, non-volatile, or a combination of volatile and non-volatile.

[0060] Here, "or" is inclusive and non-exclusive unless explicitly stated otherwise or indicated by the context. Thus, "A or B" here means "A, B, or both" unless otherwise explicitly indicated or indicated by the context. Furthermore, "and" is joint and various unless otherwise explicitly indicated or indicated by the context. Thus, "A and B" herein means "A and B, jointly or individually" unless otherwise explicitly indicated or indicated by the context.

[0061] The scope of this disclosure includes all changes, substitutions, variations, modifications, and alterations to the exemplary embodiments described or illustrated herein as understood by those skilled in the art. The scope of this disclosure is not limited to the exemplary embodiments described or illustrated herein. Furthermore, while this disclosure describes and illustrates each embodiment herein as comprising a specific component, element, feature, function, operation, or step, any of these embodiments may include any combination or permutation of any component, element, feature, function, operation, or step described or illustrated anywhere in this specification as understood by those skilled in the art. Additionally, references in the appended claims to a device or system or component that is suitablely configured, arranged, and operated to perform a specific function or is operable include such device, system, or component, whether such device, system, or component is satisfied, configured, arranged, operated, or operable, or whether a specific function is enabled or disabled. Furthermore, while this disclosure describes or illustrates specific embodiments as providing specific advantages, specific embodiments may not provide such advantages, or may provide some or all of them.

Claims

Claim 1 A method for detecting a tank in a railway environment comprises: identifying a tank within an image of the railway environment by means of an image detection tool; analyzing the tank to determine one or more characteristics associated with the tank by means of the image detection tool; determining a plurality of tank classifications by means of the image detection tool, wherein the plurality of tank classifications include a first tank classification associated with hazardous materials and a second tank classification associated with non-hazardous materials; classifying the tank into the first tank classification based on the one or more characteristics by means of the image detection tool; identifying a railway track within the image of the railway environment by means of the image detection tool; determining the distance between the tank and the railway track by means of the image detection tool; comparing the distance between the tank and the railway track with a predetermined threshold distance by means of the image detection tool; classifying the tank into the first tank classification; and determining whether the tank poses a risk to the railway environment by means of the image detection tool based on the step of comparing the distance between the tank and the railway track with the predetermined threshold distance. A method for detecting a tank in a railway environment, comprising the step of generating an alarm in response to the determination by the image detection tool that the tank poses a danger to the railway environment. Claim 2 A method for detecting a tank in a railway environment according to claim 1, further comprising the step of determining that the distance between the tank and the railway track is less than or equal to the predetermined threshold distance, wherein the predetermined threshold distance is 50 feet. Claim 3 A method for detecting a tank in a railway environment according to claim 1, comprising the step of classifying the risk as high risk, medium risk, or low risk based on the distance between the tank and the railway track by the image detection tool, wherein the high risk is associated with a distance between 0 and 20 feet; the medium risk is associated with a distance between 21 and 30 feet; and the low risk is associated with a distance between 31 and 50 feet. Claim 4 A method for detecting a tank in a railway environment, wherein, in claim 1, the image is captured by a camera attached to a train vehicle moving along the railway track through the railway environment; and the train vehicle is one of a locomotive; or a geometry vehicle. Claim 5 A method for detecting a tank in a railway environment according to claim 1, further comprising the step of training the image detection tool to identify the tank using a machine learning model. Claim 6 A method for detecting a tank in a railway environment according to claim 1, wherein the image comprises the railway track and one or more other railway tracks; the railway track is closer to the tank than the one or more other railway tracks; and the distance between the tank and the railway track is the shortest distance measured when viewed in a plane from the centerline of the outer rail of the railway track to the outer edge of the tank. Claim 7 A method for detecting a tank in a railway environment according to claim 1, wherein the alarm comprises: an indication of the danger; an indication of the level of danger associated with the danger; a description of the tank; the location of the tank; the distance between the tank and the railway track; an image of the railway environment; the time at which the image of the railway environment was captured; and the date on which the image of the railway environment was captured. Claim 8 In a device for detecting a tank in a railway environment, the device comprises one or more processors and a memory for storing instructions, and when executed by the one or more processors, the instructions cause the one or more processors to: identify a tank within an image of the railway environment; analyze the tank to determine one or more characteristics associated with the tank; determine a plurality of tank classifications—the plurality of tank classifications include a first tank classification associated with hazardous materials and a second tank classification associated with non-hazardous materials—; classify the tank into the first tank classification based on the one or more characteristics; identify a railway track within the image of the railway environment; determine the distance between the tank and the railway track; compare the distance between the tank and the railway track with a predetermined threshold distance; classify the tank into the first tank classification; and determine whether the tank represents a risk to the railway environment based on comparing the distance between the tank and the railway track with the predetermined threshold distance. A device for detecting a tank in a railway environment that performs an operation including generating an alarm in response to a determination that the tank indicates the danger in the railway environment. Claim 9 In claim 8, the operation further comprises determining that the distance between the tank and the railway track is less than or equal to the predetermined threshold distance, and the device for detecting the tank in a railway environment where the predetermined threshold distance is 50 feet. Claim 10 In paragraph 8, the operation further comprises classifying a risk as high risk, medium risk, or low risk based on the distance between the tank and the railway track, wherein the high risk is associated with a distance between 0 and 20 feet; the medium risk is associated with a distance between 21 and 30 feet; and the low risk is associated with a distance between 31 and 50 feet, for a device for detecting a tank in a railway environment. Claim 11 In claim 8, the image is captured by a camera attached to a train vehicle moving along the railway track through the railway environment; and the train vehicle is a device for detecting a tank in the railway environment, which is one of a locomotive or a geometry vehicle. Claim 12 In claim 8, the above operation further comprises training an image detection tool to identify the tank using a machine learning model, for a device for detecting a tank in a railway environment. Claim 13 A device for detecting a tank in a railway environment, wherein, in claim 8, the image comprises the railway track and one or more other railway tracks; the railway track is closer to the tank than the one or more other railway tracks; and the distance between the tank and the railway track is the shortest distance measured when viewed in a plane from the centerline of the outer rail of the railway track to the outer edge of the tank. Claim 14 In claim 8, the alarm comprises at least one of: an indication of the danger; an indication of the level of danger associated with the danger; a description of the tank; the location of the tank; the distance between the tank and the railway track; an image of the railway environment; the time at which the image of the railway environment was captured; and the date at which the image of the railway environment was captured. Claim 15 In one or more computer-readable storage media containing instructions for detecting a tank in a railway environment, when executed by a processor, the instructions cause the processor to: identify a tank within an image of the railway environment; analyze the tank to determine one or more characteristics associated with the tank; determine a plurality of tank classifications—the plurality of tank classifications including a first tank classification associated with hazardous materials and a second tank classification associated with non-hazardous materials—; classify the tank into the first tank classification based on the one or more characteristics; identify a railway track within the image of the railway environment; determine the distance between the tank and the railway track; compare the distance between the tank and the railway track with a predetermined threshold distance; classify the tank into the first tank classification; and determine whether the tank poses a risk to the railway environment based on comparing the distance between the tank and the railway track with the predetermined threshold distance; One or more computer-readable storage media containing a command for detecting a tank in a railway environment, which enables the operation including generating an alarm in response to a determination that the tank indicates the danger in the railway environment. Claim 16 In paragraph 15, the operation further comprises determining that the distance between the tank and the railway track is less than or equal to the predetermined threshold distance, and one or more computer-readable storage media containing a command for detecting the tank in a railway environment where the predetermined threshold distance is 50 feet. Claim 17 In paragraph 15, the operation further comprises classifying the risk as high risk, medium risk, or low risk based on the distance between the tank and the railway track, wherein the high risk is associated with a distance between 0 and 20 feet; the medium risk is associated with a distance between 21 and 30 feet; and the low risk is associated with a distance between 31 and 50 feet, one or more computer-readable storage media containing a command for detecting a tank in a railway environment. Claim 18 In paragraph 15, the image is captured by a camera attached to a train vehicle moving along the railway track through the railway environment; and the train vehicle is one of a locomotive; or one or more computer-readable storage media containing a command for detecting a tank in the railway environment. Claim 19 In paragraph 15, the above operation comprises one or more computer-readable storage media containing commands for detecting tanks in a railway environment, which further includes training an image detection tool to identify the tank using a machine learning model. Claim 20 In paragraph 15, the image comprises the railway track and one or more other railway tracks; the railway track is closer to the tank than the one or more other railway tracks; and one or more computer-readable storage media containing a command for detecting a tank in a railway environment, wherein the distance between the tank and the railway track is the shortest distance measured when viewed in a plane from the centerline of the outer rail of the railway track to the outer edge of the tank. Claim 21 A device for detecting a tank in a railway environment, comprising: means for identifying a tank within an image of the railway environment; means for analyzing the tank to determine one or more characteristics associated with the tank; means for determining a plurality of tank classifications—the plurality of tank classifications including a first tank classification associated with hazardous materials and a second tank classification associated with non-hazardous materials—; means for classifying the tank into the first tank classification based on the one or more characteristics; means for identifying a railway track within an image of the railway environment; means for determining a distance between the tank and the railway track; means for comparing the distance between the tank and the railway track with a predetermined threshold distance; means for classifying the tank into the first tank classification; and means for comparing the distance between the tank and the railway track with the predetermined threshold distance; means for determining whether the tank poses a danger to the railway environment based on the above; and means for generating an alarm in response to the determination that the tank poses a danger to the railway environment. Claim 22 A device for detecting a tank in a railway environment, comprising, in addition to means for implementing a method according to any one of paragraphs 2 through 7, in paragraph 21. Claim 23 A computer-readable medium containing a command for detecting a tank in a railway environment, comprising a command that causes the computer to perform a step of the method of any one of claims 1 to 7 when executed by a computer. Claim 24 A method for detecting a tank in a railway environment, comprising: determining a plurality of tank classifications by an image detection tool, wherein the plurality of tank classifications include: a first tank classification associated with a tank containing hazardous substances; and a second tank classification associated with a tank containing non-hazardous substances; identifying a tank within an image of a railway environment by the image detection tool; analyzing the tank by the image detection tool to determine one or more characteristics associated with the tank; classifying the tank into the first tank classification based on the one or more characteristics by the image detection tool; determining by the image detection tool whether the tank contains hazardous substances and whether the tank poses a risk to the railway environment based on classifying the tank into the first tank classification; and generating an alarm by the image detection tool in response to the determination that the tank poses a risk to the railway environment. Claim 25 A method for detecting a tank in a railway environment according to claim 24, further comprising: a step of determining the distance between the tank and the railway track by the image detection tool; and a step of determining by the image detection tool whether the distance between the tank and the railway track is less than or equal to a predetermined threshold distance—wherein the predetermined threshold distance is 50 feet. Claim 26 A method for detecting a tank in a railway environment according to claim 24, further comprising the step of classifying the risk as high risk, medium risk, or low risk by the image detection tool. Claim 27 In paragraph 24, the above image is a method for detecting a tank in a railway environment, which is captured by a camera attached to a train vehicle moving along a railway track through a railway environment. Claim 28 A method for detecting a tank in a railway environment according to claim 24, further comprising the step of training the image detection tool to identify the tank using a machine learning model. Claim 29 A method for detecting a tank in a railway environment, wherein one or more of the above characteristics include: the length of the tank; the width of the tank; the height of the tank relative to the ground; the shape of the tank; and the roof type of the tank. Claim 30 A method for detecting a tank in a railway environment according to claim 24, wherein the alarm comprises at least one of: the danger indication; the danger level associated with the danger; the description of the tank; the location of the tank; the image of the railway environment; the time at which the image of the railway environment was captured; and the date the railway environment was photographed. Claim 31 A device for detecting a tank in a railway environment, wherein the device comprises one or more processors and a memory including instructions, and when executed by the one or more processors, the instructions cause the one or more processors to perform operations including: determining a plurality of tank classifications, wherein the plurality of tank classifications include: a first tank classification associated with a tank containing hazardous substances; and a second tank classification associated with a tank containing non-hazardous substances; identifying a tank within an image of a railway environment; analyzing the tank to determine one or more characteristics associated with the tank; classifying the tank into the first tank classification based on the one or more characteristics; determining whether the tank contains the hazardous substance and whether the tank poses a risk to the railway environment based on classifying the tank into the first tank classification; and generating an alarm in response to the determination that the tank poses a risk to the railway environment. Claim 32 An apparatus for detecting a tank in a railway environment, wherein, in paragraph 31, the operation further comprises: determining the distance between the tank and the railway track; and determining whether the distance between the tank and the railway track is less than or equal to a predetermined threshold distance—wherein the predetermined threshold distance is 50 feet. Claim 33 In paragraph 31, the above operation further comprises classifying the risk as high risk, medium risk, or low risk, a device for detecting a tank in a railway environment. Claim 34 In paragraph 31, the above image is a device for detecting a tank in a railway environment, which is captured by a camera attached to a train vehicle moving along a railway track through a railway environment. Claim 35 In claim 31, the above operation further comprises training an image detection tool to identify the tank using a machine learning model. A device for detecting a tank in a railway environment. Claim 36 In paragraph 31, the above one or more characteristics include one or more of the following: the length of the tank; the width of the tank; the height of the tank relative to the ground; the shape of the tank; and the roof type of the tank. A device for detecting a tank in a railway environment. Claim 37 In paragraph 31, the alarm comprises at least one of: an indication of the danger; an indication of the level of danger associated with the danger; a description of the tank; the location of the tank; an image of the railway environment; the time at which the image of the railway environment was captured; and the date the railway environment was photographed. A device for detecting a tank in a railway environment. Claim 38 A non-transient computer-readable storage medium that implements a command, wherein when executed by a processor, the command causes the processor to perform operations including: determining a plurality of tank classifications, wherein the plurality of tank classifications include: a first tank classification associated with a tank containing hazardous substances; and a second tank classification associated with a tank containing non-hazardous substances; identifying a tank within an image of a railway environment; analyzing the tank to determine one or more characteristics associated with the tank; classifying the tank into the first tank classification based on the one or more characteristics; determining whether the tank contains the hazardous substance and whether the tank poses a risk to the railway environment based on classifying the tank into the first tank classification; and generating an alarm in response to the determination that the tank poses a risk to the railway environment. Claim 39 In paragraph 38, the above operation further comprises: determining the distance between the tank and the railway track; and determining whether the distance between the tank and the railway track is less than or equal to a predetermined threshold distance—wherein the predetermined threshold distance is 50 feet—a non-transient computer-readable storage medium. Claim 40 In paragraph 38, the above operation further comprises classifying the risk as high risk, medium risk, or low risk in a non-transient computer-readable storage medium. Claim 41 In paragraph 38, the image is a non-transient computer-readable storage medium captured by a camera attached to a train car moving along a railway track through the railway environment. Claim 42 In paragraph 38, the above operation further comprises a non-transient computer-readable storage medium that trains an image detection tool to identify tanks using a machine learning model. Claim 43 In paragraph 38, one or more of the above characteristics comprise: the length of the tank; the width of the tank; the height of the tank relative to the ground; the shape of the tank; and one or more of the roof type of the tank, in a non-transient computer-readable storage medium. Claim 44 A device for detecting a tank in a railway environment, comprising: means for determining a plurality of tank classifications, wherein the plurality of tank classifications include: a first tank classification associated with a tank containing hazardous substances; and a second tank classification associated with a tank containing non-hazardous substances; means for identifying a tank within an image of a railway environment; means for analyzing the tank to determine one or more characteristics associated with the tank; means for classifying the tank into the first tank classification based on the one or more characteristics; means for determining whether the tank contains the hazardous substance and whether the tank poses a risk to the railway environment based on classifying the tank into the first tank classification; and means for generating an alarm in response to the determination that the tank poses a risk to the railway environment. Claim 45 A device for detecting a tank in a railway environment, comprising, in addition to means for implementing a method according to any one of paragraphs 25 through 30, in paragraph 44. Claim 46 A computer-readable medium containing instructions for detecting a tank in a railway environment, comprising instructions that cause the computer to perform a step of the method of any one of claims 24 through 30 when executed by a computer.

Images