Technical Field

[001] The present invention relates in general to metal processing plants. More particularly, the present invention relates to controlling movement of metal product in a metal processing plant.

Background

[002] In a metal processing plant, raw material, intermediate product, or semi-finished/ finished metal product typically moves between various sections (or stages or partitions) such as rolling section, cooling section, packaging section etc. This can involve moving metal products such as bars, coils, sheets etc., over conveyors for moving the metal product(s) from one section to another section. This may additionally be done to inspect the metal product in the stage. For example, before packaging, the metal product may be inspected by operating personnel, when they are on a monitoring bay (conveyor).

[003] While the metal product is transported on such conveyors, there is movement of the metal product between sections by various conveyors. Such movement is guided and controlled by various mechanical units as stoppers, lifters etc, which are initiated by precise tracking of the movement of the material.

[004] Currently, proximity sensors or field sensors placed in the conveyors are used to detect presence of product on the conveyor or bay. These are typically devices that sense presence of the metal product such as bars or coils etc, and either a contact or non-contact type, but typically from a very close distance from the material. Accordingly, they are collocated with the conveyors (e.g. directly below or very close to or on the conveyor assembly. As the metal product is transported and moved through the conveyors, these sensors detect the presence on the conveyor.

[005] Due to ambient conditions, hot material which is to be tracked, and positioning of these devices, these sensors are prone to heavy stresses (e.g. extreme heat, pressure etc.). Also, there are possibilities that the material comes in contact with the sensor due to misalignment or poor quality of the material, making the instruments prone to failure or wrong reading. This often leads to damage or failure of the sensors causing hinderance and delay in the processing (or a part of the processing) in the metal processing plant. The failure of the sensors may lead to plant downtime which may additionally require replacement/configuring of new sensors in the conveyors.

[006] In some metal processing plants, lasers are used in place of such physical sensors. Lasers need to be positioned in large numbers and at different positions close to the conveyors. As they need to be still collocated, they are also prone to similar stresses. Many such sensors or lasers are required for monitoring the metal product and controlling the stoppers.

[007] As these devices are not mission critical (i.e. critical to the main process of producing and/or packaging metal products), alternative systems and methods are needed to remove the process dependency on these devices.

Summary

[008] The present invention relates to a control system and a method for controlling movement of metal product in a metal processing plant. The metal processing plant can be one of, but not limited to, a bar mill, a hot and cold rolling mill, a wire rod mill and the like. The metal processing plant comprises a conveyor assembly which includes a plurality of conveyors (also referred as conveyor bays or bays). The conveyor assembly receives metal product for transporting. The metal product can be transported between various sections (or parts) of the metal processing plant in different stages of the manufacturing process. For example, the metal product can be received from a cooling section for transporting to a packaging section for packaging. In an embodiment, the metal processing plant is a bar mill and the metal product are a plurality of bars. The metal product alternately can be, but not be limited to, wire rods, rolled coils, metal strips/sheets and the like. Movement of the metal product on the plurality of conveyors is controlled using one or more stoppers located at predetermined locations on the conveyor assembly.

[009] The control system comprises an input interface and a controller. The input interface receives visual feed of each conveyor of the plurality of conveyors captured by one or more cameras. Each camera of the one or more cameras is located at a predetermined position for capturing the visual feed of the conveyor. A camera may be mounted on a static location, or it may be mounted on a rail or other suitable frame, on which it can move for collecting visuals. Accordingly, the predetermined position can be static (for static location) or a variable position on the rail. The controller receives the visual feed captured by the one or more cameras and processes the visual feed with a model for image processing. The model is generated with historical and/or visual data of the conveyor assembly.

[010] In real-time, for each image frame in the visual feed, the controller obtains a plurality of values of a predetermined parameter for one or more portions of the image frame. This can involve converting the image to a binary image by applying 2D convolution and image filtering with the model. The plurality of values of the predetermined parameter are obtained using the model. The predetermined parameter is a pixel parameter. For example, the parameter can be intensity for each pixel (or pixel intensity).

[Oi l] In one embodiment, the one or more portions comprise a first contour associated with an entry of the conveyor and a second contour associated with an exit of the conveyor. Thus, for each conveyor, two contours may be defined. The contours can be polygons having three or more sides. For example, the contours can be triangles, quadrilaterals, pentagons etc. In accordance with the embodiment, the controller obtains the plurality of values for the predetermined parameter for the first contour and the second contour.

[012] Further, for each image frame, the controller compares the plurality of values of the predetermined parameter with one or more threshold values for the predetermined parameter. The model provides the one or more threshold values for the comparison. The one or more threshold values are associated with detecting presence of the metal product on the conveyor assembly.

[013] In one embodiment, the one or more threshold values are associated with detecting the presence of the metal product in the first contour and the second contour. The one or more threshold values comprise a first threshold value associated with the first contour and a second threshold value associated with the second contour. Thus, based on the comparison, the controller can determine if there is any metal product on either or both first and second contours.

[014] According to the comparison of the plurality of values with the threshold values, the controller detects presence of the metal product on at least one conveyor. Accordingly, the controller generates a signal for control of the one or more stoppers. This is based on the detection and status of the one or more stoppers. [015] Additionally, the controller can be configured to detect presence of a personnel on at least one conveyor of the conveyor assembly with the model. In such case, the controller generates the signal based on detecting the presence of the metal product, the presence of the personnel and the status of the one or more stoppers.

[016] The control system can also have a communication interface for communicating the signal to one of a Human Machine Interface (HMI) and the one or more stoppers to control operation of the one or more stoppers based on the detection.

[017] The HMI is configured to display presence of the metal product on the conveyor assembly, and status of the one or more stoppers, to receive an input for operating the one or more stoppers. Thus, the controller can generate the signal for the HMI. In an embodiment, the HMI also displays presence of personnel on the conveyor assembly.

[018] The movement of the metal product can be monitored and controlled using one or more devices. For instance, the method can be performed with the control system as described above. Alternately, the method can be performed with other suitable computing device(s) that is operationally connected with the cameras, the stoppers and that are configured to perform the method.

[019] In accordance with different embodiments, the method comprises receiving visual feed of each conveyor of the plurality of conveyors captured by the one or more cameras. The method further comprises obtaining (with the model) the plurality of values of the predetermined parameter for one or more portions of each image frame in the visual feed. In addition, the method comprises comparing, for each image frame, the plurality of values of the predetermined parameter with one or more threshold values for the predetermined parameter. Here, the model also provides the one or more threshold values for the comparison. Thereafter, the method comprises detecting presence of the metal product on at least one conveyor based on the comparison. Additionally, the method comprises generating a signal for control of the one or more stoppers based on the detection of the metal product on the at least one conveyor and status of the one or more stoppers. Brief Description of Drawings

[020] Figure 1 shows an environment of a metal processing plant which comprises a control system for controlling movement of metal product, in accordance with an embodiment of the invention;

[021 ] Figure 2 is a block diagram of the control system for controlling movement of the metal product, in accordance with an embodiment of the invention;

[022] Figures 3A and 3B are simplified representations of image frames of a conveyor with the metal product, in accordance with an embodiment of the invention;

[023] Figures 3C, 3D and 3E are simplified representations of training images, in accordance with an embodiment of the invention;

[024] Figures 4A-4C show image frames with various contours / regions, in accordance with an embodiment of the invention;

[025] Figure 4D is a simplified representation of a first contour and a second contour in an image frame, in accordance with an embodiment of the invention; and

[026] Figure 5 is a flowchart of a method for monitoring and controlling movement of the metal product in the metal processing plant, in accordance with an embodiment of the invention.

Detailed Description

[027] In industrial environments such as in the case of metal manufacturing, products in different stages of manufacturing are often transported on surfaces that are made from similar materials, or from materials having similar physical properties. Examples of these include metal on metal, paper web on polymeric surfaces, plastic on plastic etc. In such cases, it is difficult to differentiate the product from the underlying surface.

[028] Consider the case of metal bars in a metal processing plant. Manufacturing of these bars may involve receiving a huge bar from a blast furnace and providing this to one or multiple rollers (or other suitable equipment) where it is converted into thinner and longer bars. This may also involve cutting these rolled or drawn bar(s) (or rod(s)) into multiple thinner bars or rods, or performing other operations as would be apparent to those of skill in the art.

[029] Eventually, the bars or rods or other product (in case of metal plant) gets cooled in a cooling section. From the cooling section, the bars are moved for packaging. In the packaging section the metal product such as bars are transported (or arranged) on conveyors or packaging bays. These are metallic conveyors, and thus there is a metal product on a metal surface. The present invention provides a system and method for controlling movement of metal products in such environments.

[030] Figure 1 shows one exemplary environment 100 of a metal processing plant which comprises a control system 101 for controlling one or more equipment in the metal processing plant. The metal processing plant comprises a conveyor assembly 103, consisting of a plurality of conveyors or bays (105i, 105 ₂ , . , 105 _N, referred as plurality of conveyors 105).

There can be two or multiple conveyors (or bays) according to the scale of the plant. Also, bays may be arranged serially or in parallel, or there can be a combination of series and parallel bays, and such arrangements of conveyors in metal plants will be apparent to those skilled in the art.

[031] During operation, the conveyor assembly 103 receives metal product 109 (which in Fig. 1 includes a plurality of bars (e.g. steel bars)). This can be for receiving the metal product from a cooling section 113 of the metal processing plant for packaging in a packaging section. The metal product may alternately be transported between other sections (as per the requirement), and the transport between the cooling and packaging sections is one possible requirement. The conveyor assembly can be part of the packaging section or can simply move the metal bars for packaging to another dedicated section for packaging. The packaging involves inspecting quality of the metal product (bars) and transferring approved metal product 109 for packaging.

[032] In Fig. 1, the conveyor assembly 103 receives the plurality of bars at a first conveyor 1051 and initiates movement of the plurality of bars to either another conveyor (next or last available conveyor) or to a packaging bay for bundling the plurality of bars together. For controlling the movement of the plurality of bars, the conveyor assembly 103 has one or more stoppers ( 1071 , 1072 _, . . . , 107 _N ) which are placed at predetermined locations. A stopper has a metal body (e.g. a rectangular frame) and is typically provided at entry or exit of a conveyor. Such a stopper can be raised or lowered to control the movement of the metal product 109 from or to the conveyor. For example, a stopper at the exit of the first bay is lowered when the plurality of bars are initiated to move from the first bay to the second bay. These actions need to be performed in real-time. The stopper operation is linked with the movement of the metal product on the conveyor assembly. Optionally, personnel may want to stop the movement and remove faulty product. For example, short bars may need to be removed. In such a case, the stopper operation is also linked with safety of the personnel and the stoppers are to be kept raised till the personnel have performed the inspection and moved away from the conveyor or bay.

[033] Movement of the metal product can be monitored with visual inspection. Visual inspection may additionally involve monitoring the bays for presence of personnel. For visual inspection, the metal processing plant comprises a plurality of cameras (l l li,

1112. 11 IN, also referred as one or more cameras 111). Each camera of the one or more cameras 111 is located at a predefined position for monitoring a corresponding conveyor of the plurality of conveyors 105.

[034] The position of each camera is near to the conveyor monitored by it, such that images of the conveyor is captured to detect the presence of the metal product 109 on the conveyor (or for other image analytics). Thus, the location of a camera for monitoring a bay is such that it can take a visual feed (e.g. image or video) of the conveyor in real-time. In other words, the camera is not located on the conveyor and the functionality of the camera does not affect the process (meaning that the process can also work without the camera feed). A camera may be mounted on a static location, or it may be mounted on a rail or other suitable frame, on which it can move for collecting visuals. Accordingly, the predetermined position can be static (for static location) or a variable position on the rail.

[035] The position (predefined position(s)) of the one or more cameras 111 may be decided based on various parameters. Examples of these include, but are not limited to, lighting conditions in the metal processing plant, camera capability and settings such as zoom factor, or environmental conditions such as, reflective surfaces and the like. Accordingly, after considering all the parameters (or few critical parameters), the one or more cameras 111 can be mounted on a wall, bridge or any other position(s) near to the plurality of conveyors 105. This is to ensure that the cameras can capture the visual feed of the conveyor. As compared to proximity sensors or lasers, which may be close to the product within few millimetres, the cameras would typically be few meters away from the conveyor. In an embodiment, one camera of the one or more cameras 111 is dedicated for one conveyor of the plurality of conveyors 105. In another embodiment, more than one camera may be dedicated for each conveyor. This depends on factors such as the product, the desired accuracy and the plant environment. For instance, some products might need different views of the product to be able to accurately detect the product on the conveyor as they are moved on the conveyors.

[036] Thus, each camera(s) captures visual feed of a respective conveyor during the process operation in the metal processing plant. The visual feed may include a live sequence of images of conveyors. In an embodiment, the one or more cameras 111 may record signals which are encoded to image format. The visual feed of the plurality of conveyors 105 captured by the one or more cameras 111 is received by the control system 101.

[037] The control system 101 is configured with a model (not shown in Figure 1, covered in Figure 2). This model is an image processing model that is generated offline based on sample visual data of the conveyor assembly. Image data of each conveyor is analysed for generating the model.

[038] The model can be generated using various techniques such as using one or more of machine learning, filtering and so forth. In an embodiment, the model is trained using deep learning techniques such as Convolutional Neural Networks (CNN) with historic visual data of conveyor assembly 103. The historic visual data includes a plurality of images (or visuals) of conveyors. This can have variations such as different light conditions, empty conveyors, conveyor occupied with metal product etc.

[039] The training of the model is explained hereinbelow.

[040] The plurality of images of conveyors occupied with metal product 109 is provided for training the model. Figures 3A and 3B are simplified representations of image frames of a conveyor with the metal product 109. Figure 3A represents a top view of the conveyor, wherein the metal product 109 is fully received on the conveyor. Figure 3B represents an intermediate stage, wherein the metal product 109 is partly received on the conveyor. As would be apparent, there are different states of the metal product on the conveyor. There can be different number of metal product (e.g. less or more bars or rolls), different lighting conditions, camera orientation etc. To appropriately identify various image aspects when the metal product is present on the conveyor, different images in different conditions are needed.

[041] Additionally, the segregation of pixel values between the conveyors 105 and the metal product 109 is provided for training. The segregation of the pixel values may also be provided with a help of an area boundary. For example, the segregation may be provided such that pixel range outside the area boundary is associated with the conveyor and pixel range inside the area boundary is not defined. The model is trained using statistical distribution such as, Gaussian Mixture Models (GMMs) to model the pixel range associated with the conveyor and the metal product. A person skilled in the art would understand that any other technique for modelling the pixel values, not mentioned explicitly may also be used in the present invention.

[042] Thus, based on the GMM technique, unknown pixel values within the area boundary in the plurality of images are assigned. Essentially, the segmentation for pixel values in the plurality of images may be learned based on intensity (e.g. colour) of pixel values segmented previously. Once the segregation (segmentation) of pixel values is learned, a shape of the conveyor is represented using contours.

[043] Initial contours may be obtained by fitting an area boundary such as a quadrilateral which may circumscribe the conveyor. Figure 3C shows a simplified representation of an image frame used during training, wherein the conveyor is fitted with a quadrilateral for representing a contour region. The contour region 301 in Figure 3C is represented by a dotted line. Further, the model is trained to identify a region of interest (ROI) in the plurality of images, where the ROI represents the conveyor.

[044] The ROI may be obtained by approximating the initial area using a polygonal approximation technique. Figure 3D represents an exemplary image frame, wherein the contour region represented with the dotted line is approximated using polygonal technique. The approximated polygon region 303 is represented with a highlight (solid) line.

[045] The images of the conveyor assembly with and without the metal product are used to appropriately detect the presence of metal product. In Figures 3A - 3D, the images provide reference of foreground - i.e. when metal product is present on metal background. [046] Additionally, the model is trained for determining one or more portions on empty conveyors. These portions can assist in efficiently separating the foreground from background in real-time. Conveyors can have different surfaces. There can be metal sheets that are combined to form the conveyor. This can lead to there being different textures on the conveyor. For example, there can be gaps between metal plates such as gap 305. Gaps are generally hollow and therefore are not metallic in appearance.

[047] Presence of such different portions (gaps) is helpful in capturing presence of the metal product in operation. Thus, the model is trained using images of unloaded conveyors. Figure 3E is a simplified representation of an image frame of an unloaded conveyor. The one or more portions are identified from such image frames. These portions can be polygonal, circular or other geometric contours. The model is trained using Otsu, gaussian mean, KNN, or canny edge detection etc., for determining the one or more portions.

[048] As mentioned, the one or more portions can be polygonal. As an example, one such polygon is a four-sided polygon such as quadrilateral, rectangle etc. The one or more portions may vary in shape, size depending on type of ROI, lighting condition in the metal processing plant, view angle of the cameras and the like.

[049] In an embodiment, the model is trained to determine contours associated with entry and exit in the conveyor. The historic images (history data or historic visual data) of unloaded conveyors is segmented using thresholding techniques such as, Otsu, gaussian mean, KNN, canny edge detection and the like. Figure 4A shows a sample output of an image (image frame) of a conveyor, which is segmented using one such thresholding technique. The segmented images may be operated using bitwise OR functions on the conveyor image, as shown in Figure 4B. The images operated at Figure 4B are processed by applying convolution techniques such as, dilation, inversion, morphology operations and the like., as shown in Figure 4C. From the convoluted images, candidate contours are obtained by joining all continuous values having same colour or intensity with respect to geometric and shape, as shown in Figure 4C.

[050] Thus, in accordance with the embodiment, the model is trained with pixel values of one or more contours obtained such as, front contour 401 and rear contour 403 as shown in Figure 4C, which represent entry and exit of conveyors in the images. It is to be noted that the shape / size of the contours need not be limited to the quadrilateral as shown in Figure. 4. These can be selected according to the conveyor structure / surfaces, and there may be different number of portions or contours identified for generating the model.

[051] According to the contours and the pixel intensity values, the model is trained for determining one or more threshold values for the contour regions. The threshold values may be learned from historic visual data, which may be taken over different times of day and in different lighting / plant conditions. The threshold values may be adopted dynamically based on environmental conditions in the metal processing plant.

[052] For example, consider a plurality of unoccupied and occupied conveyor images represented by,

E = ]ei] and O = ]oi], i = 1...N.

[053] In the above, E represents an empty conveyor, while O represents a conveyor fully occupied with metal product.

[054] The first contour values and second contour values are obtained for all the empty and occupied conveyor images. Further, a model, for example, a gaussian model is assigned for both the first contour and the second contour values separately for each unoccupied and occupied conveyor image. Thereafter, the threshold values may be selected as the values which may separate the two (first contour and second contour) models. In other words, different contours can have different threshold values. These may be intensity values (or other suitable pixel property) for separating the metal product and the conveyor.

[055] In addition, the model is trained to identify presence of personnel on the conveyors. Images with presence of operators can be used for this training. Thus, the model can also detect presence of operating personnel on the conveyor assembly.

[056] In an embodiment, the model is used to compute for each pixel distribution of intensities to identify static and moving pixels. The pixel distribution for various frames is combined with existing status of the bay and the status of person detection (e.g. in memory for last 20 / 25 frames) to compute the presence of the person on the conveyor in the current frame. The model can be trained based on image data with operator(s) on conveyor. [057] The trained model is used by the control system 101 in real-time for controlling the movement of the metal product 109 on the conveyor assembly.

[058] Figure 2 is a simplified block diagram of the control system 101, in accordance with an embodiment of the present invention. As shown, the control system 101 comprises an input interface 201, a model 202, a controller 203 and a communication interface 204.

[059] The input interface may include I/O’s of the control system 101, which connect the control system 101 with the one or more cameras 111. Thus, the control system 101 communicates with other components through the I/O’s. The control system 101 receives the visual feed from the one or more cameras 111 through the I/O’s. The visual feed received from the one or more cameras 111 is associated with the plurality of conveyors 105 captured during the process operation in the metal processing plant.

[060] The control system 101 is configured with the model 202, which has been trained (refer description of Figure 1 above) for the metal processing plant.

[061] The controller 203 is configured to process the visual feed received from the one or more cameras 111. On receiving the visual feed of the conveyor, the controller 203 may convert images from the visual feed into binary images using two-dimensional convolution techniques. The binary images include two possible values for each pixel in the images, i.e., black (represented as“1”) and white (represented as”0”).

[062] Thus, for each binary image (image or image frame), the controller 203 obtain a plurality of values of a predetermined parameter for one or more portions. In an embodiment, the predetermined parameter is intensity for each pixel in each portion of the binary image. Thus, the controller 203 obtains intensity value for each pixel in the one or more portions of the binary image. The intensity value indicates either black (“1”) or white (“0”). The controller 203 obtains the plurality of values for the one or more portions using the model 202 which is trained for obtaining the plurality of values.

[063] In an embodiment, the one or more portions include a first contour associated with an entry of the conveyor and a second contour associated with an exit of the conveyor. The controller 203 using the model 202 obtains the plurality of values of intensity for the first contour and the second contour. Figure 4D is a simplified representation of an image frame of a conveyor where the two contours are shown. The first contour 405 is near the entry of the conveyor and the second contour 407 is near the exit of the conveyor.

[064] The pixel intensity for the pixels in the first contour and the second contour is compared with one or more threshold values associated with intensity. Thus, the pixel intensity values for first contour are compared with the threshold values for the first contour, and the pixel intensity values for the second contour are compared with the threshold values for second contour. As described above, the one or more threshold values are associated with detecting presence of the metal product 109 on the conveyor assembly and is provided by the model 202.

[065] As there can be differences in condition, there can be different thresholds. The one or more threshold values may vary based on shape of contour, lighting condition, presence of reflective surfaces and the like. In an embodiment, the one or more threshold values include a first threshold value associated with the first contour and a second threshold value associated with the second contour.

[066] In an embodiment, the comparison of the intensity values may involve checking a ratio of black and white pixels. These can be obtained by either comparing intensity against the one or more threshold values or by having the binary image. For example, if the intensity value is greater than a threshold, the pixel can be said as black, otherwise the pixel is marked white. Or in case of a binary image 1 means black pixel and 0 means white pixel. For the various portions, a ratio of black pixels to white pixels can provide if there is metal product on the portion or if the portion is empty (i.e. empty part of conveyor). For example, the metal product 109 is detected, if the value of black pixel is more than white pixel in one portion, (for example, 60 (black): 40 (white)).

[067] Consider the first contour 405 and the second contour 407 in the Figure.4E. The pixel values of black and white intensity in the first contour 405 is compared with the first threshold value and the pixel values of black and white intensity in the second contour 407 is compared with the second threshold value to detect the presence of the metal product 109. For example, if the value of black intensity is more than the one or more threshold values, the metal product 109 is detected on the conveyors. [068] Thereafter, based on the detection of the metal product 109 and status of one or more stoppers 107, the controller 203 generates a signal which can be used for controlling the one or more stoppers 107. [069] Table 1 shows an exemplary scenario of checking status and generating signals for controlling one or more stoppers, based on the comparison of the plurality of values of pixel intensity with the one or more thresholds.

Table 1

[070] In Table 1, CPF and CPR represent the first contour and second contour values, the parameters (a, b) represents the first threshold value and the second threshold value. The status shows the presence of metal product on the conveyor. [071] As mentioned, operating personnel may be assessing the quality of the metal product.

In case, if a defected product is identified, the personnel may require removing the defected product before packaging. In such cases, the controller 203 may additionally detect presence of the personnel from the visual feed on at least one conveyor of the conveyor assembly 103 using the model 202. Thus, in such cases, the controller 203 generates the signal based on the detection of the metal product 109, the presence of the personnel and the status of the one or more stoppers 107.

[072] The signal can be communicated by the controller to other devices. The control system can have a communication interface, which is connected to a Human Machine Interface (HMI) of the plant. The HMI can encompass both software and hardware components required for exchanging information. The HMI may include, for example, touch display panels, mobile devices, computers devices and the like. Thus, the controller can send the signal to an HMI that can be used by the operator to check where the metal product is present on the conveyor assembly, or if there are any personnel on the conveyor assembly. Along with this the status of the stoppers on the conveyors is also displayed. This can assist the operator in taking appropriate action.

[073] The communication interface may alternately connect the controller with the one or more stoppers, and the signal can be sent by the controller to one or more stoppers. For instance, consider the metal product 109 is detected at the entry of the first conveyor 105i in the conveyor assembly 103 and subsequent conveyor in the conveyor assembly is determined to be unoccupied. In such case, the stopper associated with the first conveyor 1051 may be provided with a signal to be lowered to allow the metal product 109 to move further to the subsequent conveyor for processing.

[074] The above Figures 1 - 4 are explained considering the metal processing plant as the bar mill and the metal product 109 to be plurality of bars. The present invention is not restricted to bar mills or to metal bars. The system can also be implemented in other metal processing plants such as hot or cold rolling mills, wire rod mills etc. Additionally, the present invention can also be implemented in other industries such as plastic and paper industries, where there is such a foreground / background image analytics required for controlling industrial processes.

[075] Referring now to Figure 5, which is a flowchart of a method for controlling movement of metal product in a metal processing plant, in accordance with an embodiment of the present invention. Various steps of the method may be performed by the control system 101, or at least in part by the control system 101. The method may alternately be implemented with PLCs or other suitable computing devices that can be configured to perform the steps of the method.

[076] At 501, the visual feed of each conveyor of the plurality of conveyors 105 is received (e.g. by the controller 203). The visual feed of the plurality of conveyors 105 is captured by the one or more cameras 111. The visual feed is processed (e.g. by the controller) to obtain the plurality of binary image frames with the model (using the 2D convolution technique and image filtering).

[077] At 502, the plurality of values of the predetermined parameter such as pixel intensity is obtained for one or more portions in each image frame. For example, the values are obtained for two portions, where the first portion is associated with the first contour and the second portion is associated with the second contour. The plurality of values may be obtained using the model 202, which is trained for detecting the first contour and the second contour and for obtaining the values of pixel intensity (or other suitable pixel parameters).

[078] At 503, the plurality of values of the pixel parameter (e.g. intensity) are compared with the one or more threshold values associated with the pixel parameter. For example, the pixel intensity values obtained for the first contour are compared with the first threshold value associated with the first contour and the pixel intensity values obtained for the second contour are compared with the second threshold value.

[079] At 504, the presence of the metal product 109 on at least one conveyor is detected based on the comparison. As mentioned, the one or more threshold values can be associated with detecting presence of the metal product 109 on the conveyor assembly 103. Here, the controller 203 may compare the plurality of pixel intensity values with the threshold values. In an embodiment, the controller 203 checks the ratio of black and white intensity values against the one or more threshold values for detecting presence of the metal product 109 on the conveyor. For example, the metal product 109 is detected, if the value of black pixel is more than white pixel in one portion, (for example, 60 (black):40 (white)). Thus, ratio of the intensity values of white and black pixel in both the first contour and the second contour is compared to detect the presence of the metal product 109.

[080] At 505, a signal for controlling of the one or more stoppers 107 is generated. The signal for controlling the one or more stoppers 107 may be generated based on output of the comparison. For example, consider the metal product 109 is in the first conveyor and the second conveyor is busy. In such case, the stopper associated with the movement of the metal product 109 to the second conveyor is signalled to be raised to stop the movement of the metal product 109 to the second conveyor. Alternately the signal is for an HMI. Thus, if the metal product 109 is present, the HMI may display the presence of the metal product 109 on the conveyor assembly 103.

[081] In some embodiments, the method also includes detecting presence of personnel. In such case, the control is based on the presence of metal product, the presence of personnel and status of the stoppers.

[082] As the method disclosed herein is non-invasive, i.e. the sensing is performed using cameras that are not located on the conveyors, but away from the conveyors, the present invention eliminates need to have contact-based sensors/ proximity sensors. Also, the same system can be used to detect human presence, and thus provides added utility.

[083] The present invention enables remote inspection of the metal processing plant, which can assist in improving safety, quality and productivity of the process.

[084] The present invention provides a digitally enabled solution for controlling process operation in metal processing plant, which can be extended to other process industries such as, plastic and paper industries.

REFERRAL NUMERALS