Description:FIELD OF INVENTION
The present invention refers to a video surveillance system and method in general. In particular, the present invention relates to a real time video surveillance system for detection and tracking of cash theft based on CCTV vision and AI detection and method thereof.

BACKGROUND OF THE INVENTION
Cash theft by cashiers is a persistent and costly issue in retail environments. Traditional security measures, such as CCTV systems, only record video footage and require manual review after an incident has occurred. Further, these systems and technology also do not provide real-time detection or alerts for suspicious cash handling, making it difficult for business owners to prevent losses as they happen.

Cash theft by cashiers has a significant and multifaceted impact on retail businesses. Financially, it directly reduces profits and can accumulate into substantial losses over time, sometimes even pushing companies toward financial distress or bankruptcy. Beyond the immediate monetary loss, employee theft undermines workplace morale and trust, as honest staff may feel demoralized or unfairly scrutinized due to increased security measures. The discovery of such theft can also damage a company's reputation, leading to a loss of customer trust and potentially driving customers away. Additionally, dealing with theft diverts management attention and resources from core business activities, further hampering productivity and growth. In severe cases, the fallout can include legal action, increased insurance costs, and stricter internal controls, all of which add to operational burdens and costs.

Existing solutions rely heavily on human observation or basic security protocols, which are often reactive and limited in scope. As a result, theft or misappropriation of cash may go unnoticed until after significant losses have occurred.
The existing technology for cash theft detection are as follows:
US20190378389A1 relates to a fraud-detecting computer-based system which includes
a barcode scanner;
an image capture device configured to receive video data from the cashier barcode scanning area;
a memory configured to store the data;
a data processing module configured to receive and to process the video data signals from an image capture device and to receive and process the signals from the barcode scanner.
US’389 discloses that in order to perform the fraud detection, while the video data processing, the data processing module detects the timing of placing an item against the barcode scanner and automatically compares it with the timing of receiving a signal from the barcode scanner. It further discloses that once cashier fraud is detected, a notification to a user of a computer system is sent.

US20030135406A1 discloses a computer-implemented method and computer program for use with a cash register that tracks individual transaction data by employee. The computer implemented method further helps to identify unusual patterns that are indicative of employee theft. US’406 discloses the method of detecting cash register employee theft comprising:
allowing the tracking of certain transaction data by individual employee,
thereafter, comparing that data to other employees and/or the same employee time shifts from different days,
then, if an unusual or suspicious patterns of activity can be identified and thereafter investigated by the employer to determine whether dishonest activities are occurring.

CN110472870A relates to a Cashier Service Regulation Detection System Based on Artificial Intelligence. CN’870 discloses the cashier service regulation detection system based on artificial intelligence which includes-
1) Image Acquisition camera to monitor and acquire cash register region scene in real time Face, and transmit video flowing;
2) employee's situation identifies server to detect staff and customer situation in real-time detection super market checkout area;
3) medium detection service device to detect the positioning function of the work related objective medium on cashier Energy;
4) check-out services detector to detect the check of the staff and customer with legitimate bill;
5) violation affair alarm device to send phonetic alarm and short message to the server, when recognizing employee has violation operation, Real-time performing audio alert or the supervisory terminal for sending alarm text information and pictorial information to administrative staff.
CN107426535A relates to an anti-lost processing system for video of supermarket check out counters goods including:
a camera installing zone is provided with video frequency pick-up head, and the camera is used in monitor and detection area;
a video processing area is transferred to scanning area close to scanning area, goods after video processing domain scanning;
a detection zone to detect the object when it enter inside that particular area;
a scanning area to scan the video as processed from the video processed area; and
a warning system to send alarm under abnormal conditions;

There is also no system/technology/techniques currently available that can automatically notify business owners or authorities immediately upon detecting theft or suspicious activity related to cash theft during real time activity. Therefore, there is a need to provide system/technology/techniques automatically notify business owners or authorities immediately upon detecting theft or suspicious activity related to cash theft during real time activity. So, the retail shops/company/bank/organization may take immediate action while there is cash theft is happening or any person is trying to do cash theft.
It is an object of the present invention to addresses these gaps by introducing a system and method/technique that leverages computer vision and artificial intelligence to monitor cashier activity through existing CCTV cameras. Further, at the same time the present invention provides the accurate result for any cash theft.
The present invention detects and tracks the handling of cash, recognize patterns indicative of theft (such as pocketing or hiding cash), and provide real-time alerts to the management/organization/security surveillance system. This proactive approach reduces reliance on manual monitoring, increases detection accuracy, and helps prevent losses before they escalate. The same approach is also able to detect the real time recognition of the cash theft by the cashier in the retail environments.

SUMMARY OF THE INVENTION
The present invention refers to a video surveillance system and method in general. In particular, the present invention relates to a real time video surveillance system for accurate detection and tracking of cash theft and method thereof.

The real time video surveillance system for accurate detection and tracking of cash theft comprises:
• a memory configured to store the information therein; and
• a processor configured to access a plurality of video frames captured using one or more overhead video cameras installed in an organization.

The processor is configured to extract one or more images of the space of the organization from the plurality of video frames. Further, the processor is configured to process the one or more videos to identify cash as present in the hand of the person as present in the space. Each detection corresponds to the cash as present in the hand of the person. In addition, the processor is configured to analyze the detections of the cash to be placed by that particular person to define at least one of true positives and false positives associated with movement of person’s hand by placement of the cash by that particular person within the space. Furthermore, the processor is configured to maintain a check of placement of the cash by that particular person based upon at least one of the true positives and the false positives.

According to yet another embodiment, a method for detecting and tracking cash theft by a person in an organization. The method includes
• accessing a plurality of video frames captured using one or more overhead video cameras installed in a space of an organization;
• processing the one or more videos to identify cash as present in hand of a person in the space;
• each detection corresponds to a cash present in hand of a person as present in the space of the organization;
• analyzing the detections to define at least one of true positives and false positives associated with movement of cash in hand of that particular person as present within the space.

The real time video surveillance system and method for detection and tracking of cash theft ensures high accuracy, continuous analysis, reduce false positive alarm, and enhances security and loss of cash prevention in retail environments or any organization or any bank. Additionally, the feedback loop enables the real time c detection of cash theft cash theft detection and tracking system over time, adapting to new cash theft behaviours and minimizing errors.

The summary is provided to introduce the system as a representative concept in a simplified form that is further described below in the detailed description. This summary is not intended to limit the key essential features of the present invention nor its scope and application.
Other objectives, features and advantages of the enclosed embodiments will be apparent from the following detailed disclosure, from the attached dependent claims as well as from the drawings.
Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise herein. All references to "a/an/the element, apparatus, component, means, module, step, etc." are to be interpreted openly as referring to at least one instance of the element, apparatus, component, means, module, step, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not have to be performed in the exact order disclosed, unless explicitly stated.

BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWINGS
Embodiments are described with reference to the following Figures. The same numbers may be used throughout to reference like features and components that are shown in the Figures:
Figure 1 illustrates a cash theft detection and tracking system implemented according to the aspects of the present invention.
Figure 2 illustrates a real time video surveillance system for cash theft in accordance with the present invention.
Figure 3 illustrates an example a method for tracking of detections using the cash theft detection and tracking of FIG. 1, implemented according to the aspects of present technique in accordance with the present invention.
Figure 4 illustrates an example process for training the detection and tracking module for learning for result using the system of FIG. 1, implemented according to the aspects of present technique.

The present invention can be understood with reference to the detailed figures and description set forth herein. Various embodiments are discussed below with reference to the figures. However, those skilled in the art will readily appreciate that the detailed descriptions given herein with respect to the figures are simply for an explanation of the invention as the methods and systems may extend beyond the described embodiments. For example, the teachings presented and the needs of a particular application yield multiple alternatives and suitable approaches to implement the functionality of any detail described herein. Therefore, any approach extends beyond the particular implementation choices in the following embodiments described and shown.

References to “one embodiment,” “at least one embodiment,” “an embodiment,” “one example,” “an example,” “for example,” and so on indicate that the embodiment(s) or example(s) may include a particular feature, structure, circuit, architecture, characteristic, property, element, or limitation but that not every embodiment or example necessarily includes that particular feature, circuit, architecture, structure, characteristic, property, element, or limitation. Further, repeated use of the phrase “in an embodiment” does not necessarily refer to the same embodiment.

DESCRIPTION:
The embodiments herein and the various features and advantageous details thereof are explained more comprehensively with reference to the non-limiting embodiments that are detailed in the following description. Descriptions of well-known components and processing techniques are omitted so as to not unnecessarily obscure the embodiments herein.
The following description includes the preferred best mode of one embodiment of the present invention. It will be clear from this description of the invention that the invention is not limited to these illustrated embodiments but that the invention also includes a variety of modifications and embodiments thereto. Therefore, the present description should be seen as illustrative and not limiting.
Unless otherwise specified, all terms used in disclosing the invention, including technical and scientific terms, have the meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. By means of further guidance, term definitions may be included to better appreciate the teaching of the present invention.
As used in the description herein, the meaning of "a", "an" and "the" includes plural reference unless the context clearly dictates otherwise. Also, as used in the description herein, the meaning of "in" includes "in" and "on" unless the context clearly dictates otherwise.
Before discussing example embodiments in more detail, it is noted that some example embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations as sequential processes and in the forms of steps to perform the operation, many of the steps may be performed in parallel, concurrently or simultaneously. In addition, the order of operations may be re - arranged. The processes may be terminated when their operations are completed but may also have additional steps not included in the figure. The processes may correspond to methods, functions, procedures, subroutines, steps, etc.
Specific structural and functional details disclosed herein are merely representative for purposes of describing example embodiments of the present invention. Inventive concepts may, however, be embodied in many alternate or optional forms and should not be taken as limited to only the example embodiments set forth herein.
At least one example embodiment is generally directed to detection of people in a space of the organization. In particular, the present techniques discloses a specific technique to detect the cash theft by the cashier on a real time basis.
The present invention relates to a real time video surveillance system for accurate detection of cash theft and method thereof.

The real time video surveillance system for accurate detection of cash theft comprises:
• a memory configured to store the information therein; and
• a processor configured to access a plurality of video frames captured using one or more overhead video cameras installed in an organization.
The processor is configured to extract one or more images of the space of the organization from the plurality of video frames. Further, the processor is configured to process the one or more images or videos to identify cash as present in the hand of the person as present in the space. Each detection corresponds to the cash as present in the hand of the person. In addition, the processor is configured to analyze the detections of the cash to be placed by that particular person to define at least one of true positives and false positives associated with movement of person’s hand by placement of the cash by that particular person within the space. Furthermore, the processor is configured to maintain a check of placement of the cash by that particular person based upon at least one of the true positives and the false positives.

Figure 1 illustrates an example cash theft detection and tracking system 100 for real time cash theft detection and sending alert to the management team of the organization to stop the theft activity using the present technique. The cash theft detection and tracking system 100 includes a memory 105, a processor 106 and a plurality of CCTV video cameras installed within the space of the organization. As will be appreciated by one skilled in the art the cash theft detection and tracking system 100 may be used for real time cash theft detection and sending alert to the organization which may present in a variety of spaces such as a retail store, a hospital, a bank, etc. Each of the CCTV video camera is configured to capture real - time videos at a pre - determined resolution corresponding to pre - determined detection scales.
Moreover, a communication module 102 is coupled with the processor 106. The communication module 102 is configured to access a plurality of video frames such as generated on a real time basis. In some examples, the video frames of the CCTV cameras may be stored in the memory 105 and the processor 106 is configured to access the video frames from the memory 105. Further, the memory 105 also stores the cash theft activity as confirmed by the organization and the suspicious activity.
In the illustrated embodiment, the processor 106 includes an image extractor 111, a processing module 112 and a tracking module 117. In this embodiment, the processing module 112 includes an activity based detector 116 which detects the activity of the cashier while the cashier is counting the cash.
In an embodiment the data processing module could be a CPU, microprocessor, ECM (electronic computing machine), PLC (programmable logic controller), or an integrated circuit configured to execute particular data processing commands (instructions, applications).
Further, the memory is configured to store data and is also stated as storage module and could be, but not limited to, a hard-drive disk (HDD), flash-drive, ROM (read-only memory), solid-state drive (SSD), etc.
In the embodiment of the present invention, the CCTV video camera is installed at the place where the activity (101) of the cashier can be easily detected. The CCTV video camera send the video frames to the cash theft detection system (100) using the communication module (102). The communication module (102) is connected to the processor (106) to process the cash theft activity or suspicious activity by the cashier. Thereafter, the communication module (102) will send the real time video frames of the cashier activity to the image extractor (111) which will extract the video frames of the cashier when there is cash in hand. The image extractor (111) is configured to extract one or more video frames from the CCTV camera as placed near the cashier. Further, the image extractor (111) is also configured to get the real time video frames from the CCTV camera as placed near cashier. The processor (106) will get activated once cash is detected in the hand of the cashier. The image extractor (111) will then extract the video frames and will forward the same to the processing module (112). The processing module (112) includes the activity based detector (116) which is configured to process one or more video frames and check the activity of the cashier when cash is present in his hand. The activity based detector (116) detects the activity of the cahier when cash is present in his hand. It will detect that after getting the cash, the cashier is keeping the cash at the pre-specified place or drawer or other places which is not specified by the organization to keep the cash.
If the activity based detector (116) detects any unauthorised activity or not defined activity, then it will treat the same as the suspicious activity and then it will send an alert or notification to security surveillance system. The alert will be sent to the management using the alert module (120) and then a final output (121) will be sent. Further, if the cashier is keeping the cash in his bag or is pocket then the activity based detector (116) will detect the activity as the cash theft and then also an alert and a notification will be sent to security surveillance system. The alert will be sent to the management using the alert module (120) and then a final output (121) will be sent.
The tracking module (117) is configured to analyze the detections to define at least one of true positives and false positives associated with movement of cash as present in the hand of the cashier. Moreover, the tracking module (117) is configured to maintain a count of the cash as present in the hand of cashier based upon at least the one of the true positives and false positives. In an embodiment, the tracking module (117) is configured to analyze the detections to define missed detections of the cash. The tracking module (117) is configured to maintain a substantially accurate count of the cash as present in the hand of the casshier, true positives and false positives results.

Additionally, this real time cash theft detection system is configured so that once cashier fraud is detected, an alert or notification is sent to a defined user, for example, to a security service employee. Further, the alert or notification will be also sent to the security service employee when there is any suspicious activity so that the management team of the organization is able to stop the cash theft if the cashier is trying to do any illegitimate activity. Also, the present real time cash theft detection and tracking system is able to detect the suspicious and cash theft activity in a real time, so that the management is able to stop the cash theft activity by the cashier or another person.
The alert or notification to the user could be an SMS or MMS or an email with a report about the analysis attached. This configuration helps the organization to promptly respond to the detection of a fraud/theft or the suspicious activity, as well as to take action against the responsible person.

Figure 2 illustrates the block diagram of the real time video surveillance system for detection of cash theft using a model applied to real-time CCTV footage (100) in accordance with the present invention. The cash theft detection and tracking system (100) comprises of a video extraction module (201) configured to receive CCTV footage from a retail store. This CCTV footage is then provided as input to a pre-processing module (203) that utilizes a batch processor with a large buffer to ensure real time frame processing without drops.

According to one embodiment of the present invention, the cash theft detection and tracking system (100) comprises:
• a plurality of video extraction module (201) to receive continuous video inputs from CCTV cameras installed in a retail store;
• a plurality of preprocessing module (203) to handle the video frame processing efficiently using a batch processor;
• a plurality of detection module (205) to detect the cash theft activity by cashier/any person;
• a plurality of alert module (120) to send alert when any suspicious activity is detected; and
• a plurality of storage module (105) for retaining flagged video footage for later review and verification; and
The CCTV as present in the organization/retail shop/bank provide the video input to the video extraction module (201) and then the extracted video frames will be pre-processed using the preprocessing module (203). The preprocessing module (203) will thus handle the video frame processing efficiently using a batch processor.

The pre-processed video frame data will then be stored in the storage module (105) which may be a memory to store all the data related to the cash theft and is in direct communication with the processor (106). Thereafter, the detected video frames will be analysed using the hand movement of every person and the cashier with the objects, pockets and bags using fine-tuned object detection neural network. The object detection neural network may be YOLO module or SAM module or SAM2 module or any other module which is capable to detect the object. Then the detection module (205) may employ a fine tuned YOLO module to detect the hand movement on a real-time and very accurately and then send the alert of any suspicious activity to the store-keeper or management team of the organization or to the security surveillance system in a very less span of time. The detection module (205) is also connected to the storage module (105) in order to store all the details are performed by detection module (205) for future functioning of the real time video surveillance system (100). The alert module (120) send alert when any suspicious activity is detected. The suspicious and normal activity as detected using the finetuned YOLO module is stored in the storage module (105) for retaining flagged video footage for later review and verification.
The real time cash theft detection and tracking system (100) also includes a specific batch processor with a large buffer size which ensures that every frame (of usually 3-5 second duration) is processed without drops, enabling real-time operation. The batch processor processes up to 30 streams concurrently of the RTSP Feed received by input video camera feed in order to perform video surveillance related to large buffer size. The real time cash theft detection and tracking system (100) also incorporates a learning feedback loop, where employee verifications refine the model, reducing false positives over time.
The real time cash theft detection and tracking system (100) includes the detection module (205) which is trained on a custom dataset, including variations in store layouts, lighting, occlusions, and crowd density, to minimize false positives and enhance accuracy

According to the one embodiment of the present invention, the real time cash theft detection and tracking system (100) includes the storage module (105) which retains flagged video footage for later review. As a result, the real time cash theft detection and tracking system (100) is capable to review the video frame data as stored in the storage module (105) to later review and analyse the video data.
In this embodiment of the present invention, the detection module (205) is configured on a custom dataset, including variations in store layouts, lighting, occlusions, and crowd density, to minimize false positives and enhance accuracy. Further, the detection module (205) continuously analyzes the video feed in real-time using fine-tuned object detection neural network module such as YOLO module rather than predefined rules. The preprocessing module (203) prevents frame drops using a batch processor with a large buffer size to maintain real-time analysis without data loss. The flagged videos are stored by the storage module (105) for security and evidence purposes and can be reviewed by authorized personnel. The employees or security surveillance system can manually verify and classify flagged incidents as cash theft or non-cash theft using an integrated dashboard interface.

According to yet another embodiment, a method for detecting and tracking cash theft by a person in an organization. The method includes
• accessing a plurality of video frames captured using one or more overhead video cameras installed in a space of an organization;
• processing the one or more images to identify cash as present in a person in the space;
• each detection corresponds to a cash present in hand of a person as available in the space of the organization;
• analyzing the detections to define at least one of true positives and false positives associated with movement of cash in hand of that particular person as present within the space.

Figure 3 illustrates an example a method for tracking of detections using the cash theft detection and tracking of FIG. 1, implemented according to the aspects of present technique in accordance with the present invention. The method involved for detection of cash theft based on AI detection using the real time cash theft detection and tracking system in accordance with the present invention. At step 301, the video extraction module (201) receives CCTV footage from a retail store or bank or any place or any organization. So, the real time video frames are pushed from the CCTV camera in the form of video footage. Then at step 302, this CCTV footage is provided as input to a pre-processing module (203) that utilizes a batch processor with a large buffer to ensure real time frame processing without drops. Then at step 303, person localization of the output from pre-processed module is done using any object detection neural network model such as YOLO model description. The fine tuned YOLO model is used to detect the cash theft activity by any cashier or any person. The YOLO techniques is fined tuned in such a manner that it can be customised as per the user. The activity of the person carrying the cash can be defined and that defined activity can be forwarded to the detection module (205) to detect the cash theft activity and the suspicious activity. So, the extracting feature values related to hand movements of the person with cash and interactions with objects, pockets, and bags will be detected.
Then at step 304, the detection module (205) employees a fine-tuned object detection neural network model such as YOLO model to analyze hand movements and interactions of the person when that person is carrying cash in his hand and as present in the shop/organization/bank with objects, pockets, and bags. Further, the detection module (205) is accomplished on a custom dataset, including variations in store layouts, bank, organization lighting, occlusions, and crowd density, to minimize false positives and enhance accuracy.
Thereafter, at step 305, the activity of the person carrying the cash is checked while keeping the cash that where the person is keeping the cash. The detection module (205) classifies the activity into three categories: the activity of the person if classified as normal if the confidence level of the activity is less than 50%, the activity is classified as suspicious if the confidence level of the activity is between 50% to 75% and the activity is classified as cash theft if the confidence level of the activity is more than 75%. In cases of suspicious activity and at step 306, the alert module 120 will send the alert for review and verification. Then at step 307, the alert module (120) shares flagged incidents for verification using user feedback. Then at step 308, the storage module (105) retains flagged video footage for later review.
Thus, the method for real time video surveillance for detection of cash theft based on AI detection using a real time cash theft detection and tracking system detects the normal, suspicious and cash theft activity in any shop very easily, accurately and fast. As a result, the shopkeeper is able to detect the cash theft activity in real time and in within 3-5 seconds so, that the suspicious person is not able to left the shop and the cash theft activity recognises and can be stopped.

Thus, an example of a particular embodiment of a detection method for cashier fraud is described below. Figure 3 shows a flow-chart of an embodiment method for detecting cashier fraud or theft.
Further, Figure 4 illustrates an example process for training the detection and tracking module for learning for result using the system of FIG. 1, implemented according to the aspects of present technique.
The process for training the cash theft detection and tracking system for learning for result using the system is as follows:
at step 401, receiving video input data from a store’s surveillance camera;
then at step 402, pre-processing the video frames (each frame being of 3-5 seconds) and then cash object is detected when any person or cashier takes cash in his hand;
then at step 403, the detection module (205) tracks the hand movement of the person carrying the cash in hand and then extracting feature values related to hand movements and interactions with objects, pockets, and bags;
further at step 404, the fine tuned object detection neural network check the placement of the cash object by the cashier or person carrying the cash in hand;
further, at step 405, the fine tuned object detection neural network checks that the cash is placed in the cash box or not;
• in one instance and in step 406, when the cash is placed in the cash box, then it will be detected as the legitimate action and no further action will be taken; and in another instance, then at step 407, if the detection module detects that the cash is not kept in the cash box, then it will detected as an suspicious activity and cash theft activity and then a notification or alert will be sent to the security surveillance system at step 408.

Further, after detection of the suspicious activity and/or the cash theft activity, the alert module (120) will send an alert or notification to the management team or security surveillance system or organisation to take an action against the activity.
This method is implemented by a computer system with at least one memory and processor which is capable to be modified or improvement as per the requirement. The method has the stages where:
the video signals including a person with cash in hand are received and processed;
then, the video streams from the CCTV is received and detected using the detection module and the detection module is configured to receive the video data from the cashier area are received and processed;
further, during the video data processing, at least one data processing module detects at least one event of placing an item against the cash kept after its usages,
thereafter, it compares the event of receiving the signal from the cashier having cash in hand and the detected events of placing the cash by a time parameter;
if the activity do no match between the specified events is found, then at least one data processing device detects cashier fraud or suspicious activity.
Further, these methods can be embodied by a computer system, thus it can be expanded and enhanced using the embodiments which have already been described above to apply the computer system to detect cashier fraud or theft or any suspicious activity on a real time basis.

Further, other configurations may be envisaged using the cash theft detection and tracking system and method.
Thus, the present invention offers the following technical improvements and advantages:
A.) Real Time Analysis:
The present invention provides an intelligent, automated system for detecting cash theft in any organization using real-time video analysis. The batch processor with large buffer size ensures that every frame is processed without drops, enabling real-time operation.
B) Accuracy:
The system ensures high accuracy through continuous analysis and allows human verification to reduce false positives, enhancing security and loss prevention in retail environments.
C) Self Correction Model:
The feedback loop enables the model to improve over time, adapting to new cash theft and suspicious activity behaviours and minimizing errors.
As a result, the present the real time cash theft detection and tracking system (100) includes an AI retail surveillance solution which help with smart inventory management, thereby detecting product shrinkages early. The real time cash theft detection and tracking system (100) can easily integrate with ERP systems and detect any reduction in inventory levels that do not have matching purchase transactions in the records within a specified time-period. Thus, whenever there is a cash theft attempt, the present AI retail security system alerts the store manager even before the cashier/thief has actually left the store.
The present real time cash theft detection and tracking system (100) includes AI store theft mitigation systems which constitute fine-tuned AI models that are specifically trained for anomaly detection. Such fine-tuned models can identify discrepancies in counts of SKUs (stock-keeping unit’s) in the shelf versus the actual counts recorded in the database. Further, it also detects suspicious user behaviour, such as trying to replace the original cash with the counterfeit.
The capability of the real time cash theft detection and tracking system (100) is not just limited to cash theft.
Further, the present AI-powered intelligent retail video surveillance system and method scan frames within video feeds for enhanced detection of cash theft and prevention of the same. So, the present the real time cash theft detection and tracking system (100) also act as AI theft prevention systems which employ a combination of the parameters such as advanced facial recognition, behavioural analysis, hand movement, SKU identification and shelf-status detection, shelf to point-of-sales path tracking, integration with inventory database, multi-device input-feed combination, to accurately identify cash theft incidents. In addition, the present real time cash theft detection and tracking system (100) also saves the point-of-sale billing recording that can be accessed in the future for search and look up.
The real time cash theft detection and tracking system (100) including the AI theft prevention systems are arranged to trigger real-time notifications based on customisable alert thresholds and threat sensitivity levels. Further, high priority notifications can be set up for specific high-value sections of the store, such as where jewellery or mobile items are displayed. All such alerts and their timestamps, along with other incident report data, are captured and documented in the incident-database for future reference and legal purposes.
Further, using the present the real time video surveillance system (100), AI theft prevention alert thresholds can be customised according to the AI cash theft solution’s sensitivity, and the probability and frequency of occurrence of cash theft incidents at the particular organization.
It should be understood that any of the embodiments of the present system can be implemented by using hardware or by use of combination of hardware and software.
Based on the disclosure and teaching provided herein, a person of ordinary skill in the art will know and appreciate other ways and/or methods to implement embodiments of the present invention using specialized processors. Further, any of the methods describes herein may be totally or partially performed using a computer, including one or more processors, which is configured to perform the steps described herein above. Thus, embodiments are directed towards computer system including specific components to perform specific steps of any of the methods described herein above. Additionally, any of the steps of any of the methods can be performed using specific circuits.
For better understanding, aspects of the invention are described in terms of sequences of steps/arrangements that can be performed by, for example, components of a programmable computer system. It will be recognized that various steps could be performed by specialized circuits (e.g., distinct logic gates interconnected to perform a specialized function or application-specific integrated circuits), by list of steps executed by one or more processors, or by a combination of both.

While the present disclosure has been described with reference to certain embodiments and exemplary embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted without departing from the scope of the present disclosure. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the present disclosure without departing from its scope.
, Claims:We Claim:
1. A real time video surveillance system (100) for accurate detection and tracking of cash theft comprises:
a memory (105) configured to store the information therein; and
a processor (106) configured to access a plurality of video frames captured using one or more overhead video cameras installed in an organization;
wherein, the processor (106) is configured to
process the one or more videos to identify cash as present in the hand of the person as present in the space,
analyze the detections of the cash to be placed by that particular person to define at least one of true positives and false positives associated with movement of person’s hand by placement of the cash by that particular person within a particular space;
maintain a check of placement of the cash by that particular person based upon at least one of the true positives and the false positives; and
provide real time detection of the cashier theft or any suspicious activity related to the cash
2. The real time video surveillance system for accurate detection and tracking of cash theft as claimed in claim 1, wherein the processor 106 includes an video/image extractor 111, a processing module 112 and a tracking module 117.
3. The real time video surveillance system for accurate detection and tracking of cash theft as claimed in claim 2, wherein the processing module 112 comprises an activity based detector 116 which detects the real time activity of the cashier while the cashier is counting the cash.
4. The real time video surveillance system for accurate detection and tracking of cash theft as claimed in claim 3, wherein the activity based detector (116) detects the real time activity of the cahier when cash is present in his hand and also detect the activity that after getting the cash, the cashier is keeping the cash at the pre-specified place or not.
5. The real time video surveillance system for accurate detection and tracking of cash theft as claimed in claim 3, wherein the activity based detector (116) detects any un-authorised activity or not defined activity, as the suspicious activity and sends an alert or notification to the organization on a real time basis.
6. The real time video surveillance system for accurate detection and tracking of cash theft as claimed in claim 2, wherein the tracking module (117) is configured to analyze the detections to define at least one of true positives and false positives associated with movement of cash as present in the hand of the cashier on a real time.
7. The real time video surveillance system for accurate detection and tracking of cash theft as claimed in claim 6, wherein the tracking module (117) is configured to maintain a count of the cash as present in the hand of cashier based upon at least the one of the true positives and false positives.
8. The real time video surveillance system for accurate detection and tracking of cash theft as claimed in claim 6, wherein the tracking module (117) is configured to maintain a substantially accurate count of the cash as present in the hand of the cashier, true positives and false positives results.
9. The real time video surveillance system (100) for accurate detection and tracking of cash theft as claimed in claim 1, wherein, the processor processes a large buffer size to ensures that every frame of 3-5 seconds duration are processed without drops and enables real-time operation.
10. A real time video surveillance system (100) for accurate detection and tracking of cash theft comprises:
a plurality of video extraction module (201) to receive continuous video inputs from CCTV cameras installed in a retail store;
a plurality of pre-processing module (203) to handle the video frame processing efficiently using a batch processor;
a plurality of detection module (205) to detect the cash theft activity by cashier/any person;
a plurality of alert module (120) to send alert when any suspicious or theft activity is detected; and
a plurality of storage module (105) for retaining flagged video footage for later review and verification.
11. The real time video surveillance system (100) for accurate detection and tracking of cash theft as claimed in claim 10, wherein, the pre-processing module (203) handles the video frame processing efficiently using a batch processor.
12. The real time video surveillance system (100) for accurate detection and tracking of cash theft as claimed in claim 10, wherein, the detection module (205) is configured to employ a fine tuned object detection neural network to detect the hand movement on a real-time and accurate detection of any suspicious activity to an organization.
13. The real time video surveillance system (100) for accurate detection and tracking of cash theft as claimed in claim 12, wherein, the fine-tuned object detection neural network is fine-tuned YOLO model.
14. The real time video surveillance system (100) for accurate detection and tracking of cash theft as claimed in claim 10, wherein, the detection module (205) is configured to employ on a custom dataset, including variations in store layouts, lighting, occlusions, and crowd density, to minimize false positives and enhance accuracy.
15. The real time video surveillance system (100) for accurate detection and tracking of cash theft as claimed in claim 11, wherein, the batch processor processes a large buffer size to ensures that every frame of 3-5 seconds duration are processed without drops and enables real-time operation.
16. A method for real time detecting and tracking cash theft by a person in an organization, the method comprises:
• accessing a plurality of video frames captured using one or more overhead video cameras installed in a space of an organization;
• processing the one or more videos to identify cash as present in hand of a person in the space;
• each detection corresponds to a cash present in hand of a person as present in the space of the organization;
• analyzing the detections to define at least one of true positives and false positives associated with movement of cash in hand of that particular person as present within the space; and
• sending notification to the organization in case of suspicious and/or theft activity.
17. The method for real time detecting and tracking cash theft by a person in an organization as claimed in claim 16, wherein, the video frames are real time video data from a store’s surveillance camera (301).
18. The method for real time detecting and tracking cash theft by a person in an organization as claimed in claim 16, wherein, processing of video frames is initially pre-processed by a pre-processing module (302).
19. The method for real time detecting and tracking cash theft by a person in an organization as claimed in claim 16, wherein, analyzing the detected of cash in hand, along with movement of cash in hand by a detection module (205).
20. The method for real time detecting and tracking cash theft by a person in an organization as claimed in claim 16, wherein, the detection is performed by a fine tuned object detection neural network.
21. The method for real time detecting and tracking cash theft by a person in an organization as claimed in claim 20, wherein, the fine-tuned object detection neural network is fine-tuned YOLO model.
22. The method for real time detecting and tracking cash theft by a person in an organization as claimed in claim 21, wherein, the fine-tuned YOLO model tracks the activity of detected cash as present in cashier’s hand on a real time.
23. The method for real time detecting and tracking cash theft by a person in an organization as claimed in claim 21, wherein, the fine-tuned YOLO model tracks the activity of cashier’s hand that where the cash is placed.
24. The method for real time detecting and tracking cash theft by a person in an organization as claimed in claim 21, wherein, the fine-tuned YOLO model checks the activity of cashier while placement of cash from the hand within 3-5 seconds.