Description:SYSTEM AND METHOD FOR EXTRACTING DATA FROM A FINANCIAL INSTRUMENT

FIELD OF INVENTION
[0001] The embodiments of the present disclosure generally relate to a field of image capturing and processing, and specifically to a system and a method for capturing an image of a financial instrument and processing the image to extract data.

BACKGROUND OF INVENTION
[0002] Typically, banking and financial institutions face significant challenges in manually extracting essential information from cheques, such as payee name, amount, date, and cheque number. This makes the process of extracting data from the cheques time-consuming, error-prone, and inefficient. Further, variability in cheque data is a major issue, as layouts, field placements (e.g., date boxes, Indian Financial System Code (IFSC), Magnetic Ink Character Recognition (MICR)), and formatting styles (e.g., handwritten versus printed text, single versus multi line entries) differ by region or by branch. Also, presence of watermarks and background noise complicate data extraction from the cheques.
[0003] In addition, performing manual data extraction from the cheques results in high labor costs while simultaneously increasing likelihood of human errors and mistakes that lead to delays, financial discrepancies, and reprocessing. Further, processing times of the cheques are also slowed by manual verification thereby negatively impacting service efficiency. Furthermore, due to variations in handwriting, cheque formats, and physical conditions (e.g., smudges, faded ink) accuracy of data extraction from the cheques is inconsistent and requires additional verification. Finally, manual processing of the cheques lack scalability, making it difficult to handle higher transaction volumes as demand grows.
[0004] Alongside challenges of manual data extraction, automatic extraction from digital cheques also encounters several obstacles. Poor image quality in scanned or photographed cheques, often due to blurriness, shadows, or watermarks, can disrupt operations of an Optical Character Recognition (OCR) engine, resulting in errors or missed information. These variations limit effectiveness of rule-based or template-based approaches thus creating operational drawbacks.
[0005] There is, therefore, a need in the art to provide an improved system and a method to process scanned or digital cheques using dynamic image processing mechanisms to overcome the above mentioned limitations.

OBJECTS OF THE INVENTION
[0006] Some of the objects of the present disclosure, which at least one embodiment herein satisfies are listed herein below.
[0007] It is an object of the present disclosure to provide a system and a method to capture an image of a financial instrument and process the image to extract data.
[0008] It is an object of the present disclosure to provide a system and a method to process scanned or digital cheques using image processing mechanisms, machine learning mechanisms, and an Optical Character Recognition (OCR) engine.
[0009] It is an object of the present disclosure to provide a system and a method for using a skew correction mechanism, a noise removal mechanism, an object detection model, and a validation logic to accurately extract and validate cheque information.
[0010] It is an object of the present disclosure to provide a system and a method that facilitates providing a manual fallback mechanism to ensure reliable data extraction when predictions from an automated model are uncertain.

SUMMARY OF THE INVENTION
[0011] In an aspect, the present disclosure relates to a method for extracting data from a financial instrument. The method may include processing, by one or more processors of a computing device, an image of the financial instrument to obtain a plurality of image samples. The method may analyze each of the plurality of image samples to identify at least one of an entity of the financial instrument, where at least one of the entity is identified using a trained dataset. Further, the method may validate at least one of the identified entity of the financial instrument using predefined patterns. In addition, in response to correct validation of at least one of the identified entity, the method may extract data of at least one of the entity of the financial instrument.
[0012] In an embodiment, responsive to an incorrect validation of at least one of the identified entity, the data extraction from at least one of the entity of the financial instrument may include a manual validation of at least one of the entity,
[0013] In an embodiment, the image of the financial instrument may be processed using any of a skew correction mechanism and a noise removal mechanism.
[0014] In an embodiment, the trained dataset may be used to detect and locate at least one of the entity of the financial instrument using a bounding box mechanism.
[0015] In an embodiment, an OCR engine may be used to identify at least one of the entity.
[0016] In an embodiment, a cell detection mechanism may be used to extract the data of at least one of the entity, where the data is situated within cell boundaries.
[0017] In an embodiment, the cell detection mechanism may suppress the cell boundaries to extract the data of at least one of the entity.
[0018] In an embodiment, the trained dataset comprises the plurality of image samples, where each of the plurality of image samples are categorized into at least one of a training subset, a validation subset, and a test subset.
[0019] In an embodiment, a performance metrics may be applied on the trained dataset to track an impact of the training subset on the trained dataset.
[0020] In an embodiment, the test subset may be validated using the trained dataset.
[0021] In an aspect, the present disclosure relates to a system for extracting data from a financial instrument. The system may include one or more processors associated with a computing device, a memory operatively coupled to the one or more processors, wherein the memory comprises processor-executable instructions, which on execution, cause the one or more processors to process an image of the financial instrument to obtain a plurality of image samples. The one or more processors may be configured to analyze each of the plurality of image samples to identify at least one of an entity of the financial instrument. At least one of the entity may be identified using a trained dataset. Further, the one or more processors may validate at least one of the identified entity of the financial instrument using predefined patterns. Furthermore, the one or more processors may be configured to extract data of at least one of the entity of the financial instrument, in response to correct validation of at least one of the identified entity.
[0022] In an aspect, the present disclosure relates to non-transitory computer-readable medium comprising processor-executable instructions that may cause a processor to process an image of the financial instrument to obtain a plurality of image samples. The one or more processors may be configured to analyze each of the plurality of image samples to identify at least one of an entity of the financial instrument. At least one of the entity may be identified using a trained dataset. Further, the one or more processors may validate at least one of the identified entity of the financial instrument using predefined patterns. Furthermore, the one or more processors may be configured to extract data of at least one of the entity of the financial instrument, in response to correct validation of at least one of the identified entity.

BRIEF DESCRIPTION OF DRAWINGS
[0023] The accompanying drawings, which are incorporated herein, and constitute a part of this disclosure, illustrate exemplary embodiments of the disclosed methods and systems which like reference numerals refer to the same parts throughout the different drawings.
[0024] FIG. 1 illustrates an exemplary block diagram representation of a network architecture implementing a proposed system for extracting data from a financial instrument, in accordance with an embodiment of the present disclosure.
[0025] FIG. 2 illustrates exemplary functional units of the proposed system, in accordance with an embodiment of the present disclosure.
[0026] FIG. 3 illustrates an exemplary user interface of a cheque to be used for data extraction, in accordance with an embodiment of the present disclosure.
[0027] FIG. 4A illustrates an exemplary image of a date field of the cheque to be extracted, in accordance with an embodiment of the present disclosure.
[0028] FIG. 4B illustrates an exemplary formout image of the date field of the cheque to be extracted using a cell detection algorithm by suppressing cell boundaries, in accordance with an embodiment of the present disclosure.
[0029] FIG. 5 illustrates a series of steps for training a trained dataset, in accordance with an embodiment of the present disclosure.
[0030] FIG. 6 illustrates a series of steps for using the trained dataset to digitally process and extract data from the cheque, in accordance with an embodiment of the present disclosure.
[0031] FIG. 7 is a flow diagram depicting a proposed method for extracting data from the financial instrument, in accordance with an embodiment of the present disclosure.
[0032] FIG. 8 illustrates an exemplary computer system in which or with which embodiments of the present disclosure may be utilized in accordance with embodiments of the present disclosure.
[0033] The foregoing shall be more apparent from the following more detailed description of the disclosure.

DETAILED DESCRIPTION
[0034] In the following description, for the purposes of explanation, various specific details are set forth in order to provide a thorough understanding of embodiments of the present disclosure. It will be apparent, however, that embodiments of the present disclosure may be practiced without these specific details. Several features described hereafter can each be used independently of one another or with any combination of other features. An individual feature may not address all of the problems discussed above or might address only some of the problems discussed above. Some of the problems discussed above might not be fully addressed by any of the features described herein.
[0035] The word “exemplary” and/or “demonstrative” is used herein to mean serving as an example, instance, or illustration. For the avoidance of doubt, the subject matter disclosed herein is not limited by such examples. In addition, any aspect or design described herein as “exemplary” and/or “demonstrative” is not necessarily to be construed as preferred or advantageous over other aspects or designs, nor is it meant to preclude equivalent exemplary structures and techniques known to those of ordinary skill in the art. Furthermore, to the extent that the terms “includes,” “has,” “contains,” and other similar words are used in either the detailed description or the claims, such terms are intended to be inclusive in a manner similar to the term “comprising” as an open transition word without precluding any additional or other elements.
[0036] Reference throughout this specification to “one embodiment” or “an embodiment” or “an instance” or “one instance” means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present disclosure. Thus, the appearances of the phrases “in one embodiment” or “in an embodiment” in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
[0037] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. As used herein, the singular forms “a”, “an”, and “the” are intended to include the plural forms as well, unless the context indicates otherwise. It will be further understood that the terms “comprises” and/or “comprising,” when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. As used herein, the term “and/or” includes any and all combinations of one or more of the associated listed items.
[0038] Typically, banking and financial institutions face significant challenges in manually extracting key data from a financial instrument (for example, a cheque) related to such as payee names, amounts, dates, and cheque numbers. This process is often time-consuming, error-prone, and costly, with varying layouts, field placements, and formatting styles of the cheques adding complexity. Factors like watermarks, background noise, smudges, faded ink, and poor image quality further hinders accuracy, leading to delays and financial discrepancies. Additionally, manual verification slows processing and is not scalable to higher transaction volumes. Automated extraction methods, like an Optical Character Recognition (OCR) engine also struggle due to poor image quality, limiting effectiveness of rule-based or template-based approaches. The disclosed system and method facilitates dynamic and intelligent extraction of data from the financial instrument.
[0039] Various embodiments of the present disclosure will be explained in detail with reference to FIGs. 1-8.
[0040] FIG. 1 illustrates an exemplary block diagram representation of a network architecture 100 implementing a proposed financial instrument extraction system 110 (also referred to as a system 110, herewith) for extracting data from a financial instrument, according to embodiments of the present disclosure. The network architecture 100 may include the system 110, a computing device 108, a centralized server 118, a decentralized database 120. The system 110 may be communicatively connected to the centralized server 118, and the decentralized database (or node(s)) 120, via a communication network 106. The centralized server 118 may include, but is not limited to, a stand-alone server, a remote server, a cloud computing server, a dedicated server, a rack server, a server blade, a server rack, a bank of servers, a server farm, hardware supporting a part of a cloud service or system, a home server, hardware running a virtualized server, one or more processors executing code to function as a server, one or more machines performing server-side functionality as described herein, at least a portion of any of the above, some combination thereof, and the like. The communication network 106 may be a wired communication network or a wireless communication network. The wireless communication network may be any wireless communication network capable of transferring data between entities of that network such as, but is not limited to, a carrier network including a circuit-switched network, a public switched network, a Content Delivery Network (CDN) network, a Long-Term Evolution (LTE) network, a New Radio (NR), a Global System for Mobile Communications (GSM) network and a Universal Mobile Telecommunications System (UMTS) network, an Internet, intranets, Local Area Networks (LANs), Wide Area Networks (WANs), mobile communication networks, combinations thereof, and the like.
[0041] The system 110 may be implemented by way of a single device or a combination of multiple devices that may be operatively connected or networked together. For example, the system 110 may be implemented by way of a standalone device such as the centralized server 118 (and/or a decentralized server or node(s)), and the like, and may be communicatively coupled to the computing device 108. In another example, the system 110 may be implemented in/ associated with the computing device 108. In yet another example, the system 110 may be implemented in/associated with respective electronic devices 104-1, 104-2, …..., 104-N (individually referred to as electronic device 104, and collectively referred to as electronic devices 104), associated with one or more user 102-1, 102-2, …..., 102-N (individually referred to as the user 102, and collectively referred to as the users 102). In such a scenario, the system 110 may be replicated in each of the electronic devices 104. The users 102 may be a user of, but are not limited to, the financial institution, a bank, a broker, a lender, a payee, and the like. In some instances, the user 102 may include an entity or an administrator, who is in conversation with the electronic device 104. The computing device 108 may be at least one of, an electrical, an electronic, and an electromechanical device. The computing device 108 may include, but is not limited to, a mobile device, a smart- phone, a Personal Digital Assistant (PDA), a tablet computer, a phablet computer, a wearable device, a Virtual Reality/Augmented Reality (VR/AR) device, a laptop, a desktop, a server, and the like. The system 110 may be implemented in hardware or a suitable combination of hardware and software.
[0042] Further, the system 110 may include a processor 112, an Input/Output (I/O) interface 114, and a memory 116. The Input/Output (I/O) interface 114 of the system 110 may be used to receive user inputs, from one or more electronic devices 104 associated with the one or more users 102. The processor 112 may be configured to the computing device 108. The processor 112 may be coupled with the memory 116. The memory 116 may store one or more instructions that are executable by the processor to extract data from the financial instrument.
[0043] In an embodiment, the system 110 may extract data from the financial instrument. The financial instrument may be, for example, a cheque, money order, banknote, and the like. Hereinafter, the financial instrument is interchangeably referred to as the cheque for clarity. However, it is to be understood that the system 110 described herein may be readily adopted for the financial instruments of other types, such as the money orders, the banknotes, and/or the like, and configured/customized for a wide range of other applications or implementations.
[0044] For extraction of the data, an image of the financial instrument may be processed to obtain a plurality of image samples. For example, in one embodiment, the user 102 (e.g., a payee) who wants to deposit the cheque may capture multiple image samples of the cheque using the electronic device 104. The electronic device 104 may be a user image capture device, e.g., a scanner connected to a computer, a mobile device having a built-in camera, a digital camera, and/or the like. In an implementation, the user 102 may take a video clip of the cheque and submit a video file. In one embodiment, the user 102 may send the captured cheque image sample to the financial institution, e.g., a payee’s bank. In one implementation, a series of image analysis procedures may be performed on the cheque image sample to enhance the received check image sample and extract an entity of the cheque. The entity may be such as payee’s name, payee’s bank, account number, bank routing number, deposit amount, date of issue, signature, and/or the like.
[0045] In an embodiment, the image sample of the cheque may be processed using either a skew correction mechanism or a noise removal mechanism. Skew correction is a fundamental pre-processing step for the OCR engine and is primarily used for aligning text-based images. Often, text in the image sample is misaligned and so correcting this is essential to produce accurate results. The skew adjustment mechanism involves shifting one side of the image sample horizontally or vertically while keeping an opposite side fixed. Moving top or bottom of the image sample creates horizontal skew and adjusting left or right side of the image sample creates vertical skew. Further, a projection parallel to a true alignment of text lines may show highest variance. This occurs because, when the text is aligned, each projected ray through the image sample may either pass through very few black pixels (when the projected ray falls between text lines) or many black pixels (when the projected ray passes through multiple characters in a row).
[0046] Further, the image sample of the cheque may be filtered using filters to reduce noise while preserving image details. The choice of the filter may depend on behavior and type of the image sample being processed. The filtration may be done using any of a filtering mechanism, such as a weighted moving average uniform weight, a filtering mechanism with weighted moving average non-uniform weight, a weighted moving average in 2-dimensional image, and the like. In addition, the noise removal mechanism may determine geometric deformation for shake correction, in consideration of required intensity of correction and a margin amount of geometric deformation of the image sample.
[0047] In an embodiment, multiple image samples may be analyzed to identify the one or more entities that may be present on the cheque. In another embodiment, the entity of the cheque may be identified using a trained dataset. The trained dataset may be, for example, a real-time object detection system such as a You only look once (YOLO) model. In an implementation, the YOLO model uses a data acquisition and a labeling procedure that requires a varied dataset of the cheque’s image samples for training a dataset of sample images. The sample images may be sourced from publicly available datasets or collected manually. The sample images may be annotated with tools such as LabelImg that tags key fields, such as date, amount, payee name, and other entities using a bounding box mechanism available in a YOLO-compatible format.
[0048] In an embodiment, the YOLO model may include a plurality of sample images that are categorized into at least one of a training subset, a validation subset, and a test subset.
[0049] In another embodiment, a performance metrics may be applied on the YOLO model to track an impact of the training subset on the sample images of the YOLO model. The sample images of the YOLO model may be validated using the trained dataset.
[0050] In yet another embodiment, the OCR engine may be used to identify at least one of the entity.
[0051] In an embodiment, the identified one or more entities of the financial instrument may be validated using predefined patterns. Further, in response to correct validation of the identified entity, data of at least one of the entity (e.g., date) of the cheque may be extracted. In another embodiment, responsive to an incorrect validation of at least one of the identified entity, the data extraction from at least one of the entity of the financial instrument may include a manual validation of at least one of the entity.
[0052] In an embodiment, the trained YOLO model may be applied to the pre-processed images to detect, localize and locate specific entities. The YOLO model may use bounding boxes around entities, such as the date, payee name, amount, Indian Financial System Code (IFSC) code, and Magnetic Ink Character Recognition (MICR) code so as to enable precise extraction of these entities for further processing.
[0053] In an embodiment, a cell detection mechanism may be used to extract the data of at least one of the entity that is situated within cell boundaries. The cell detection mechanism may suppress the cell boundaries to extract the data of the entity. By way of an example, a cell detection and a formout processing for the entity such as a date field may be performed. For date extraction, when the cell detection mechanism is employed, a pre-existing imaging library is used to suppress cell boundaries so as to ensure that only relevant text data is available for OCR processing.
[0054] In some implementations, the system 110 may include data, and modules. As an example, the data may be stored in the memory 116 configured in the system 110. In an embodiment, the data may be stored in the memory in the form of various data structures. Additionally, the data may be organized using data models, such as relational or hierarchical data models.
[0055] In an embodiment, the data stored in the memory 116 may be processed by the modules of the system 110. The modules may be stored within the memory. In an example, the modules communicatively coupled to the processor configured in the system, may also be present outside the memory, and implemented as hardware. As used herein, the term modules refer to an Application-Specific Integrated Circuit (ASIC), an electronic circuit, a processor (shared, dedicated, or group) and the memory that execute one or more software or firmware programs, a combinational logic circuit, and/or other suitable components that provide the described functionality.
[0056] Further, the system 110 may also include other units such as a display unit, an input unit, an output unit, and the like, however the same are not shown in FIG. 1, for the purpose of clarity. Also, in FIG. 1 only a few units are shown, however, the system 110 or the network architecture 100 may include multiple such units or the system 110/network architecture 100 may include any such numbers of the units, obvious to a person skilled in the art or as required to implement the features of the present disclosure. The system 110 may be a hardware device including the processor 112 executing machine-readable program instructions to extract data from the financial instrument.
[0057] Execution of the machine-readable program instructions by the processor 112 may enable the system 110 to extract the data of at least one of the entity from the financial instrument. The “hardware” may comprise a combination of discrete components, an integrated circuit, an application-specific integrated circuit, a field-programmable gate array, a digital signal processor, or other suitable hardware. The “software” may comprise one or more objects, agents, threads, lines of code, subroutines, separate software applications, two or more lines of code, or other suitable software structures operating in one or more software applications or on one or more processors. The processor 112 may include, for example, but are not limited to, microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuits, and any devices that manipulate data or signals based on operational instructions, and the like. Among other capabilities, the processor 112 may fetch and execute computer-readable instructions in the memory 116 operationally coupled with the system 110 for performing tasks such as data processing, input/ output processing, and/or any other functions. Any reference to a task in the present disclosure may refer to an operation being or that may be performed on data.
[0058] FIG. 2 illustrates, at 200, exemplary functional units of the proposed system 110, in accordance with an exemplary embodiment of the present disclosure. The system 110 may include the one or more processor(s) 112. The one or more processor(s) 112 may be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, logic circuitries, and/or any devices that manipulate data based on operational instructions. Among other capabilities, the one or more processor(s) 112 are configured to fetch and execute computer-readable instructions stored in a memory 204. The memory 204 may store one or more computer-readable instructions or routines, which may be fetched and executed to create or share the data units over a network service. The memory 204 may include any non-transitory storage device including, for example, volatile memory such as RAM, or non-volatile memory such as EPROM, flash memory, and the like.
[0059] In an embodiment, the system 110 may also include an interface(s) 114. The interface(s) 114 may include a variety of interfaces, for example, interfaces for data input and output devices, referred to as I/O devices, storage devices, and the like. The interface(s) 114 may facilitate communication with various other devices coupled to the one or more processor(s) 112. The interface(s) 114 may also provide a communication pathway for one or more components of the one or more processor(s) 112. Examples of such components include, but are not limited to, processing engine(s) 208 and database 210.
[0060] In an embodiment, the processing engine(s) 208 may be implemented as a combination of hardware and programming (for example, programmable instructions) to implement one or more functionalities of the processing engine(s) 208. In examples described herein, such combinations of hardware and programming may be implemented in several different ways. For example, the programming for the processing engine(s) 208 may be processor executable instructions stored on a non-transitory machine-readable storage medium and the hardware for the processing engine(s) 208 may include a processing resource (for example, one or more processors), to execute such instructions.
[0061] In the present examples, the machine-readable storage medium may store instructions that, when executed by the processing resource, implement the processing engine(s) 208. In such examples, the processor(s) 112 may include the machine-readable storage medium storing the instructions and the processing resource to execute the instructions, or the machine-readable storage medium may be separate but accessible to the system 110 and the processing resource. In other examples, the processing engine(s) 208 may be implemented by electronic circuitry. The database 210 may include data that is either stored or generated as a result of functionalities implemented by any of the components of the processing engine(s) 208. In an embodiment, the processing engine(s) 208 may include a processing unit 212, an analyzing unit 214, a validating unit 216, an extracting unit 218, and other units(s) 220. The other unit(s) 220 may implement functionalities that supplement applications/ functions performed by the system 110. In another embodiment, the system 110, through the other unit(s) 220, may manage extracting data of one of the entity of the financial instrument.
[0062] The processing unit 212 may process the image of the financial instrument (e.g, the cheque) to obtain multiple image samples. The image of the financial instrument may be processed using either the skew correction mechanism and the noise removal mechanism.
[0063] The analyzing unit 214 may analyze each of the multiple image samples to identify entities of the financial instrument. The entities may be identified using the trained dataset that may be used to detect and locate the entity of the financial instrument using the bounding box mechanism. Further, an OCR engine may be used to identify at least one of the entity. The trained dataset may include the multiple image samples. The multiple image samples may be categorized into at least one of the training subset, the validation subset, and the test subset. Further, a performance metrics may be applied on the trained dataset to track an impact of the training subset on the trained dataset and the test subset may be validated using the trained dataset.
[0064] The validating unit 216 may validate the identified entities of the financial instrument using predefined patterns.
[0065] In response to the correct validation of at least one of the identified entity, the extracting unit 218 may extract data of the entities of the financial instrument. In addition, the data extraction from the entities of the financial instrument may include a manual validation of the entity, in response to the incorrect validation of the identified entity.
[0066] FIG. 3 illustrates an exemplary user interface 300 of the cheque to be used for data extraction, in accordance with an embodiment of the present disclosure. With respect to FIG. 3, fields that are commonly extracted from the cheque are, for example:
? Indian Financial System Code (IFSC): Unique 11-character alphanumeric code that identifies a specific branch of a bank in a country.
? Invoice Date: Date on which the cheque is issued.
? Account Number (A/C No.): A unique string of numbers that identifies an individual’s or organization’s bank account number.
? Amount: A payable amount mentioned on a cheque.
? Magnetic Ink Character Recognition (MICR): A 9-digit code printed at bottom of the cheque, usually next to a cheque number and the account number.
[0067] FIG. 4A illustrates an exemplary image 400 of the date field of the cheque to be extracted, in accordance with an embodiment of the present disclosure. With respect to FIG. 4A, is shown the date field that is enclosed in boundaries. FIG. 4B illustrates an exemplary formout image 450 of the date field of the cheque that is to be extracted using the cell detection algorithm by suppressing the cell boundaries, in accordance with an embodiment of the present disclosure. When data related to the date field is situated within the cells, then for extraction of the date field, the cell detection algorithm may be employed. To suppress the existing cell boundaries, the pre-existing imaging library may be used to ensure that only relevant text data is available for the OCR processing. As may be appreciated, extraction of the date field from the cheque may be made easier by using zone boundaries and identifying the cells within it. When documents are scanned into computers as images, they are stored as collections of pixels with varying colors. For analyzing such documents, an image processing mechanism is used that requires detection of boundaries and an identification of cells within the document. With advancements in deep learning, traditional convolutions have been enhanced by Convolutional Neural Networks (CNNs). Instead of manually specifying kernel filter weights, the CNNs are designed to learn them automatically from ground truth data. By leveraging hundreds of convolutions, the CNNs may effectively detect the location of the field entities of the cheque, even when the image is scanned poorly with distortion and noises.
[0068] FIG. 5 illustrates, at 500, a series of steps for training a trained dataset, in accordance with an embodiment of the present disclosure. With respect to FIG. 5, at step 502, multiple images of the financial instrument (e.g., the cheque) may be captured in a data set as data. At step 504, the multiple images may be annotated with labels and bounding boxes using various tools. The annotation may be done for various parameters of the cheque, such as date, payee name, signature, amount, and the like. At step 506, the data corresponding to the multiple images may be split into either a training set, a validation set and a test set. Next, at step 508, the YOLO model may be set up and the training set may be fed to the YOLO model, at step 510. During execution of the YOLO model, at step 512, the validation set may be used to determine performance of the model. Further, at step 514, the YOLO model may be tested using the test set. Based on test results, at step 516, the model may be adjusted and fine-tuned to prepare a trained dataset so as to accurately extract data from the financial instrument.
[0069] FIG. 6 illustrates, at 600, a series of steps for using the trained dataset to digitally process and extract data from the cheque, in accordance with an embodiment of the present disclosure. With respect to FIG. 6, multiple images of the cheque may be extracted, at step 602, using an image processing mechanism. At step 604, the YOLO model may be used to process the multiple images. After processing of the images, at step 606, mechanisms to check, for example, a date field (the entity) present on the cheque may be executed along with applying a cell detection mechanism. At step 608, data from the date field may be extracted using an OCR text mechanism. Further, the data extracted from the date field may be validated and processed, at step 610.
[0070] A set of validation rules may be applied on the entities present on the cheque. By way of an example, the entities may be validated according to predefined patterns. For instance, for:
? IFSC Code Validation: The system 110 may apply a pattern-based check, to ensure that IFSC codes follow a format of four letters, a zero, followed by six digits (e.g., ABCD0123456).
? MICR Code Validation: The system 110 may extract and verify the MICR code to meet a nine-digit requirement.
? Date Format Validation: The system 110 may validate the dates to confirm compliance with standard formats such as DD/MM/YYYY.
? Amount Formatting: The system 110 may check currency fields for appropriate formatting, based on pre-specified currency requirements.
[0071] When the validation of the entity, for example, the date is incorrect, a fallback route may be followed, at step 612. In the fallback route, the validation for the date field may be done manually. In cases where specific entities are not detected and validated accurately, the system 110 may employ a fallback logic to manually perform validation to maintain accuracy. For instance:
? To retrieve the IFSC Code, a manual search for the labels (e.g., “IFSC Code” or “IFSC”) may be performed and text adjacent to these labels may be retrieved, based on a position, so as to validate the IFSC code format.
? For MICR Code retrieval, the fallback module may invoke the pre-existing imaging library to scan the entire image and extract the MICR code manually.
? For date retrieval, if the date detection is unsuccessful, then text close to labels (e.g., “Date”) may be extracted and a cell detection technique may be applied, followed by date format validation.
[0072] At step 614, it may be ensured that the validation for all the entities present on the cheque is completed.
[0073] FIG. 7 is a flow diagram depicting a proposed method 700 for extracting data from the financial instrument, in accordance with an embodiment of the present disclosure. At step 702, the method includes processing an image of the financial instrument to obtain a plurality of image samples. The method includes analyzing, by the one or more processors, each of the plurality of image samples to identify an entity of the financial instrument, at step 704, using a trained dataset. Further, at step 706, the method includes, validating the identified entity of the financial instrument using predefined patterns. In addition, in response to correct validation of the identified entity, the method includes extracting, at step 708, data of the entity of the financial instrument. However, responsive to an incorrect validation of the identified entity, a manual validation of the entity may be performed to enable data extraction from the entity of the financial instrument.
[0074] Those skilled in the art would appreciate that embodiments of the present disclosure provides a financial instrument extraction system that employs a comprehensive workflow to ensure accurate and efficient text extraction. First, an image preprocessing module is used to correct skew, remove noise, and enhance contrast so as to optimize the financial document’s appearance for machine learning and OCR. This step improves field recognition and text extraction clarity. Further, a YOLO v7 model that is trained on a diverse dataset of images of the financial document with labeled fields, is generalized across various layouts. This is done to enable accurate detection of the entity without relying on template-based approaches. In addition, a cell detection and formout module is used to isolate and remove cell boundaries around fields like date, thereby retaining only relevant text for the OCR. This reduces visual artifacts and enhances OCR accuracy. Once fields are localized by the YOLOv7 model, the OCR may be applied to each specific region, while focusing on targeted areas to ensure precise data extraction. Field-specific validation checks may then verify an accuracy of extracted information, such as IFSC codes, MICR numbers, and date formats to correct errors and ensure compliance with required standards. Together, these steps create a robust and a reliable system for extracting data from the financial document.
[0075] FIG. 8 illustrates an exemplary computer system 800 in which or with which embodiments of the present disclosure may be implemented. As shown in FIG. 8, the computer system 800 may include an external storage device 810, a bus 820, a main memory 830, a read-only memory 840, a mass storage device 850, communication port(s) 860, and a processor 870. A person skilled in the art will appreciate that the computer system 800 may include more than one processor and communication ports. The processor 870 may include various modules associated with embodiments of the present disclosure. The communication port(s) 860 may be any of an RS-232 port for use with a modem-based dialup connection, a 10/100 Ethernet port, a Gigabit or 10 Gigabit port using copper or fiber, a serial port, a parallel port, or other existing or future ports. The communication port(s) 860 may be chosen depending on a network, such a Local Area Network (LAN), Wide Area Network (WAN), or any network to which the computer system 800 connects. The main memory 830 may be random access memory (RAM), or any other dynamic storage device commonly known in the art. The read-only memory 840 may be any static storage device(s) e.g., but not limited to, a Programmable Read Only Memory (PROM) chips for storing static information e.g., start-up or BIOS instructions for the processor 870. The mass storage device 850 may be any current or future mass storage solution, which can be used to store information and/or instructions. Exemplary mass storage device 850 includes, but is not limited to, Parallel Advanced Technology Attachment (PATA) or Serial Advanced Technology Attachment (SATA) hard disk drives or solid-state drives (internal or external, e.g., having Universal Serial Bus (USB) and/or Firewire interfaces), one or more optical discs, Redundant Array of Independent Disks (RAID) storage, e.g. an array of disks.
[0076] The bus 820 communicatively couples the processor 870 with the other memory, storage, and communication blocks. The bus 820 may be, e.g. a Peripheral Component Interconnect (PCI)/PCI Extended (PCI-X) bus, Small Computer System Interface (SCSI), USB, or the like, for connecting expansion cards, drives, and other subsystems as well as other buses, such a front side bus (FSB), which connects the processor 870 to the computer system 800.
[0077] Optionally, operator and administrative interfaces, e.g. a display, keyboard, joystick, and a cursor control device, may also be coupled to the bus 820 to support direct operator interaction with the computer system 800. Other operator and administrative interfaces can be provided through network connections connected through the communication port(s) 860. Components described above are meant only to exemplify various possibilities. In no way should the aforementioned exemplary computer system 800 limit the scope of the present disclosure.
[0078] While the foregoing describes various embodiments of the invention, other and further embodiments of the invention may be devised without departing from the basic scope thereof. The scope of the invention is determined by the claims that follow. The invention is not limited to the described embodiments, versions or examples, which are included to enable a person having ordinary skill in the art to make and use the invention when combined with information and knowledge available to the person having ordinary skill in the art.
ADVANTAGES OF THE PRESENT DISCLOSURE
[0079] The present disclosure provides a system and a method for capturing an image of a financial instrument and processing the image to extract data.
[0080] The present disclosure provides a system and a method for processing scanned or digital cheques using image processing mechanisms, machine learning mechanisms, and an Optical Character Recognition (OCR) engine.
[0081] The present disclosure provides a system and a method for using a skew correction mechanism, a noise removal mechanism, an object detection model, and a validation logic to accurately extract and validate cheque information even in poor quality images.
[0082] The present disclosure provides a system and a method that facilitates providing a manual fallback mechanism to ensure reliable data extraction when predictions from an automated model are uncertain.
[0083] The present disclosure provides a system and a method that facilitates to reduce high labor costs and remove risk of human errors associated with manual data extraction process, ensuring faster and more reliable data extraction.
[0084] The present disclosure provides a system and a method to accurately identify different fields of the cheque even when not clearly marked.
[0085] The present disclosure provides a system and a method to minimize variations in cheque designs so as to generalize and adjust to different formats thus improving scalability and reducing performance inconsistency.
[0086] The present disclosure provides a system and a method to use deep learning models for detecting fields of the cheque, combined with OCR for text extraction, for quick processing of large volumes of cheques without sacrificing accuracy or speed.
[0087] The present disclosure provides a system and a method to validate extracted fields with pattern matching and formatting rules, thus reducing need for manual corrections thereby minimizing financial discrepancies caused by incorrect data.
, Claims:
We Claim:

1. A method (700) for extracting data from a financial instrument, the method comprising:
processing (702), by one or more processors (112) of a computing device (108), an image of the financial instrument to obtain a plurality of image samples;
analyzing (704), by the one or more processors (112), each of the plurality of image samples to identify at least one of an entity of the financial instrument, where at least one of the entity is identified using a trained dataset;
validating (706), by the one or more processors (112), at least one of the identified entity of the financial instrument using predefined patterns; and
extracting (708), by the one or more processors (112), data of at least one of the entity of the financial instrument, in response to correct validation of at least one of the identified entity.
2. The method (700) as claimed in claim 1, wherein the data extraction from at least one of the entity of the financial instrument includes a manual validation of at least one of the entity, responsive to an incorrect validation of at least one of the identified entity.
3. The method (700) as claimed in claim 1, wherein the image of the financial instrument is processed using any of a skew correction mechanism and a noise removal mechanism.
4. The method (700) as claimed in claim 1, wherein the trained dataset is used to detect and locate at least one of the entity of the financial instrument using a bounding box mechanism.
5. The method (700) as claimed in claim 1, wherein an OCR engine is used to identify at least one of the entity.
6. The method (700) as claimed in claim 1, wherein a cell detection mechanism is used to extract the data of at least one of the entity, where the data is situated within cell boundaries.
7. The method (700) as claimed in claim 6, wherein the cell detection mechanism suppresses the cell boundaries to extract the data of at least one of the entity.
8. The method (700) as claimed in claim 1, wherein the trained dataset comprises the plurality of image samples, where each of the plurality of image samples are categorized into at least one of a training subset, a validation subset, and a test subset.
9. The method (700) as claimed in claim 8, wherein a performance metrics is applied on the trained dataset to track an impact of the training subset on the trained dataset.
10. The method (700) as claimed in claim 8, wherein the test subset is validated using the trained dataset.
11. A system (110) for extracting data from a financial instrument, the system comprising:
one or more processors (112) associated with a computing device (108); and
a memory (116) operatively coupled to the one or more processors (112), wherein the memory (116) comprises processor-executable instructions, which on execution, cause the one or more processors (112) to:
process an image of the financial instrument to obtain a plurality of image samples;
analyze each of the plurality of image samples to identify at least one of an entity of the financial instrument, where at least one of the entity is identified using a trained dataset;
validate at least one of the identified entity of the financial instrument using predefined patterns; and
extract data of at least one of the entity of the financial instrument, in response to correct validation of at least one of the identified entity.
12. The system (110) as claimed in claim 11, wherein the data extraction from at least one of the entity of the financial instrument includes a manual validation of at least one of the entity, responsive to an incorrect validation of at least one of the identified entity.
13. The system (110) as claimed in claim 11, wherein the image of the financial instrument is processed using any of a skew correction mechanism and a noise removal mechanism.
14. The system (110) as claimed in claim 11, wherein the trained dataset is used to detect and locate at least one of the entity of the financial instrument using a bounding box mechanism.
15. The system (110) as claimed in claim 11, wherein an OCR engine is used to identify at least one of the entity.
16. The system (110) as claimed in claim 11, wherein a cell detection mechanism is used to extract the data of at least one of the entity, where the data is situated within cell boundaries.
17. The system (110) as claimed in claim 16, wherein the cell detection mechanism suppresses the cell boundaries to extract the data of at least one of the entity.
18. The system (110) as claimed in claim 11, wherein the trained dataset comprises the plurality of image samples, where each of the plurality of image samples are categorized into at least one of a training subset, a validation subset, and a test subset.
19. The system (110) as claimed in claim 18, wherein a performance metrics is applied on the trained dataset to track an impact of the training subset on the trained dataset.
20. A non-transitory computer-readable medium comprising processor-executable instructions that cause a processor (112) to:
process an image of the financial instrument to obtain a plurality of image samples;
analyze each of the plurality of image samples to identify at least one of an entity of the financial instrument, where at least one of the entity is identified using a trained dataset;
validate at least one of the identified entity of the financial instrument using predefined patterns; and
extract data of at least one of the entity of the financial instrument, in response to correct validation of at least one of the identified entity.