Description:FIELD OF INVENTION
The present invention relates generally to the field of surveillance systems and more particularly to a smart camera system with a weatherproof enclosure coupled with edge computing hardware and a method to provide a video analytics-based solution that is intended to detect visual anomalies and deter illicit activities.

BACKGROUND OF THE INVENTION
The use of video cameras has become ubiquitous in modern times in applications related to surveillance, intrusion detection, traffic management, industrial plants, patient monitoring, etc. The cameras that are used in such applications have been improved in the contemporary era in terms of their specifications such as higher resolutions, wider angles, range of shutter speeds, autofocus, etc. However, until recent times, the cameras have provided video streams that are to be reviewed manually, either in live mode or saved footage to detect intruders, and traffic violations, confirmation of entry or exit at certain points, check on patients in distress, replay sports coverage, etc.
With the advent of Artificial Intelligence (“AI”) and Machine Learning (“ML”) models, smart cameras have evolved to a point wherein the live, or stored video footage is processed automatically to assist human intervention in processing the videos thereby increasing productivity and effectiveness. Therefore, the AI and ML models coupled to the edge computing units can automatically process the data in accordance with the use case.
For instance, the edge computing model can automatically be configured to read the number plates of the vehicles that are moving at high speeds on highways, detect traffic violations, detect processes that are not in compliance in industrial plants, or detect and raise an alert at the medical facilities when a patient is in distress by detecting the patient’s facial expressions, etc. In order to aid this automated video processing or analytics, the conventional camera systems are equipped with substantial computing power that performs effectively during the instances described above. The result of the automatically processed data is transferred to the central server for further processing in a larger context.
Therefore, the numbers that are extracted from a number plate of a speeding vehicle, and detection of intruders/trespassers in a specific premise, or the detection of patients, while they are in distress, for instance, when a patient is violently coughing, etc. can be sent to a central server, wherein memos/ challans can be generated and fines be collected for speeding vehicles, or a security officer can be alerted to take immediate action against the intruder, or a healthcare official can be alerted to attend the patient quickly. Such data processing is possible and facilitated only under normal circumstances when the camera system captures a clear and high-quality video of the object under its purview.
Whereas, in all the above instances, the effectiveness of data processing is reduced as the video camera itself is not equipped to automatically adapt to cope with sub-optimal environmental conditions while capturing, which in effect results in poor video quality that cannot be further processed by the AI/ ML models in the edge computing devices. For example, outdoor cameras that are installed for monitoring traffic and reading number plates, under adverse weather conditions transmit poor-quality videos during rainfall, or when the camera lens is misty due to moisture. Furthermore, in cases where the intruder's face is captured by the camera system and transmitted to the computing system, the face might not be clearly recognizable because of the wrong focus of the camera on the subject or if the lighting conditions are inadequate. In cases where a patient changes their position on the bed, their head or body might fall outside the viewing angle of the camera system and thus, the camera would not be able to detect the patient while they experience any pain or distress.
Automatic Number Plate Recognition (“ANPR”) is an image processing technology that captures vehicles’ images or videos and uses Optical Character Recognition (“OCR”) to read the vehicles’ number plates. Upon extraction and translation of the number plate’s information into machine-readable data, it can be used for various analytical purposes.
It can be stored and used for historical tracking, aiding law enforcement investigations, traffic studies, and security applications. ANPR systems can be integrated with databases to check for matches against stolen vehicles, inventory, wanted individuals or suspects, or even vehicles with outstanding fines. Furthermore, they can be utilized in parking lot management, automating entrance and exit processes for vehicles, vehicle inventory management, facilitating payment at fuel stations, and enhancing overall efficiency and security in various domains. Currently, there are several solutions aimed at addressing the number recognition system, which captures images of fast-moving vehicles. However, these solutions only provide an ANPR system for ideal conditions, without considering the adverse weather and environmental conditions that can disrupt the image and video quality, ultimately affecting the identification of information on number plates.
To overcome such limitations, the present invention aims to provide an intelligent visual imaging capture system equipped with a smart camera and edge computing hardware that ensures high-quality video for video analytics that is used for capturing and recognizing various images even during adverse or non-optimal environmental conditions. The present invention not only helps in capturing images and videos but also identifies any anomalies by comparing them with the existing images or videos that are readily available on the trained database. The use cases of the present invention are broad; some examples include automatic number recognition, vehicle traffic sensing technologies, face detection, intrusion detection, monitoring industrial processes (e.g., LPG bottling), patient monitoring, etc. For all such applications, the present invention provides an optimal intelligent visual capturing module that incorporates automatic positioning and training on the object of interest and optimizing for obtaining images that are blurred due to environmental conditions. It is noteworthy that the use case of the present invention is vast and is not limited to these instances.
Various prior arts have disclosed similar cameras and automatic number recognition systems:
US Patent Application US20210112183A1 discloses a camera system with waterproof features. This application is directed to a surveillance camera system that includes a magnet mount for physically receiving a camera module. The camera module in this prior art includes a housing having an exterior surface of a first shape, the surface of the magnet mount has a second shape that is substantially concave and complementary to the first shape and is configured to engage the exterior surface of the housing of the camera module. This prior art specifically only deals with waterproofing features by employing different kinds of mounts. Whereas the present invention does not only have a waterproof enclosure but is also equipped with wipers, defoggers, and dehumidifiers that can improve the humidity, and the entire system is dynamically managed and controlled at the edge and server level. The prior art provides a one-time feature, while the present invention is controlled remotely in various instances.
Chinese Patent Application CN116546065A discloses an edge gateway camera of the Internet of Things and a control method. The intention of this prior art is to gather environmental data, convert it into event data, and use this information to control IoT equipment intelligently, allowing for sensor-based automation and decision-making.
The sensing module in this prior art is used for connecting networking equipment elements within a certain range through the edge gateway of the IoT and acquiring sensing data of corresponding environments. The conversion module is used to convert the perception data into event data representing various events and the processing module converts the event data into intelligent control instructions of different scenes, controlling associated IoT equipment, and realizing sensor linkage. This prior art focuses solely on event detection and data processing, it does not enhance image quality whereas the present invention features a smart camera integrated with edge computing hardware to ensure high-quality images and videos by adjusting the camera settings. Additionally, the present invention is equipped with protective features and sensors to maintain clear video capture even in adverse weather conditions including rain, fog, poor lighting, dust, etc.
US Patent Application US11688171B2 discloses a method for person tracking and identification using intelligent camera orchestration. In this prior art, the communication interface is used to communicate with a plurality of cameras. The processor obtains metadata from an initial camera’s video stream, anticipates the future state of the object, and switches to a second camera to capture the object based on this prediction. It is pertinent to note that this prior art is all about predictive object tracking using multiple cameras whereas the present invention emphasizes visual imaging and recognition to ensure the maintenance of high image quality which is essential for applications including traffic monitoring, patient monitoring, security, surveillance, and industrial processes. Moreover, the present invention is designed to operate effectively even in unfavorable weather conditions that may otherwise degrade video quality.
US Patent Application US20210097306A1 discloses automatic license plate recognition and vehicle identification profile methods and systems that provide an automated classification of vehicle reads to build vehicle identification profiles for automated vehicle identification using content extracted from an image of a vehicle to identify the most probable vehicle identification profile. An example method in this prior art comprises capturing, by a camera, a read comprising an image frame including a portion of a vehicle; identifying at least one of a license plate number and a descriptor of the vehicle using image processing on the read; determining a probability value that the read includes the vehicle based on the identified at least one of the license plate number and the descriptor; when the probability value exceeds a threshold value, identifying a vehicle identification profile in a database using the identified at least one of the license plate number and descriptor, and updating the vehicle identification profile to include the captured read. This prior art generally provides methods and systems for utilizing image data and associated metadata to create a profile on a particular vehicle. It further provides various systems and methods identifying at least one of the characters on a license plate and descriptors (e.g., visual, or physical characteristics and features of the vehicle) from an image frame to build a vehicle identification profile. It is pertinent to note that this prior art only focuses on number recognition in ideal conditions, whereas the present invention in its preferred embodiment provides a system and method for improving the image quality with a camera equipped with external wipers, defoggers, dehumidifiers, etc., further improving the environmental lighting conditions, humidity to capture the most accurate image or video footage which is further used for analytics by leveraging machine learning and artificial intelligence models.
To solve the abovementioned drawbacks of the prior arts, the present invention discloses a smart video camera that adapts itself based on the environmental factors in which the video is captured by continuously monitoring the video quality and determining what correction actions need to be taken instantaneously in real-time in order to improve the video quality adequately enough for the AI/ ML modules to process the input data efficiently. This visual anomaly detection and number recognition system employs artificial intelligence and machine learning to continuously control the camera’s attributes, such as the focal length, aperture, and shutter speed based on the quality of the image thereby ensuring the optimal image and video quality constantly. In order to achieve the optimum image quality at all times, the system is equipped with external wipers, defoggers, and dehumidifiers, and the entire system is further housed with an IP66-rated enclosure.
In one embodiment, the present invention features live video processing in the cloud server with bookmark creation, wherein when a vehicle is captured by the camera it enables the user to easily retrieve and analyze the video clips by selecting the vehicle numbers that have been captured and identified within a specific time duration for post-mortem analysis and dispute resolution. The ML model in this system is trained with multiple input data in order to enable it to be used in various scenarios. Additionally, the custom-built ML model enables recognition of number plates of different colors, sizes, fonts, and even handwritten number plates. It features multilingual recognition for regional languages and enables recognition of the color and details of the number plate even during dark lighting conditions at night. The system can also retrieve vehicles’ registration numbers from reflective and non-reflective number plates and upon running the tests by simulating the undesirable environmental conditions, this system has shown a 100% accuracy rate in reading standard number plates. This system also allows it to be upgraded with the latest firmware with Firmware Over the Air (“FOTA”) capability, which allows for effortless wireless firmware upgrades.
Furthermore, the present system is deployed with substantial processing power that enables it to record and process high-quality video files of up to 4K resolutions (3840 x 2160) at
60 frames per second (“FPS”). The edge computation technology and the ability to store and process from local logs ensures continuous functioning of the system even in the event of internet failure or loss of connectivity to the central server. In one embodiment, the present invention can be trained to control and trigger various operations such as controlling the boom barrier based on the inputs received from the camera sensor and processing them accordingly. The system comprises a color camera with a lens having a wide recognition angle that ensures that the system captures a wide Field of View (“FOV”) of the event for accurate and efficient data processing. The present invention provides a versatile smart camera system that can be mounted on poles, or even ceilings and is highly reliable, accurate, and efficient. The system is designed for use in various applications including but not limited to parking lot management, vehicle inventory management, gas stations, entrance and exit management, and law enforcement, and is an excellent addition to any security or surveillance system.
OBJECTS OF THE INVENTION
It is the main object of the present invention to provide a smart camera and edge computing hardware system to ensure high-quality video outputs for computer vision.
It is the primary object of the present invention is to provide a visual analytics system for automatic number recognition, visual anomaly, and specificity detection.
It is another object of the present invention to provide a smart camera system that is a video analytics-based solution designed to detect, deter, and disrupt illicit activities by implementing AI and ML.
It is another object of the present invention to continuously control the camera’s attributes, such as the focal length, aperture, and shutter speed based on the image quality to ensure optimal image and video quality at all times regardless of the environmental conditions.
It is another object of the present invention to provide a camera system equipped with external wipers, defoggers, and dehumidifiers, and the entire system is housed with an
enclosure to provide enhanced image and video quality in all adverse weather conditions.
It is another object of the present invention to provide a self-optimizing, automatic, and wide camera to get the best clarity of images of the target objects.
It is another object of the present invention to optimize the camera’s attributes using the feedback of the quality of the image to get an even better image by automated control of the camera’s attributes such as the focus, aperture, shutter speed, etc.
It is another object of the present invention to provide live video processing over the cloud server and bookmark creation when a subject is captured by the camera enabling the user to retrieve and analyze the captured and processed video clips by selecting the identifier (such as the vehicle number) within a specific predefined time frame for post-mortem analysis and dispute resolution.
It is another object of the present invention to provide a custom-built AI and ML system for recognizing number plates of different colors and sizes, including private vehicles, commercial vehicles, rental vehicles, embassy or consulate vehicles, and electric vehicles.
It is another object of the present invention to provide a system that captures and recognizes number plates with different font types and sizes, as well as handwritten number plates, and features multilingual recognition for regional languages.
It is another object of the present invention to provide a camera system that can recognize the color of the number plates even during dark lighting conditions at night by altering the camera’s attributes automatically.
It is another object of the present invention to handle and retrieve vehicle numbers from both reflective and non-reflective number plates in a preferred embodiment.
It is another object of the present invention to provide FOTA capability, which allows for effortless wireless firmware upgrades.
It is another object of the present invention to provide a system with substantial processing power that enables it to record and process high-quality video files of up to 4K resolutions at 60 FPS.
It is another object of the present invention to ensure the continuous functioning of the system even in the event of internet failure or loss of connectivity to the central server by utilizing edge computation technology and local logs.
It is another object of the present invention to control and trigger various operations including but not limited to controlling the boom barrier based on the inputs received from the camera sensor and processing them based on the user requirement.
It is another object of the present invention to provide a color camera with a wide recognition angle lens to ensure that the system captures a wide FOV for accurate and efficient data processing.
It is another object of the present invention to provide a robust camera system with a weatherproof/waterproof enclosure along with wipers to ensure a clear view even during heavy rainfall and the system is further equipped with dehumidifiers to capture images in foggy conditions.
It is another object of the present invention to provide a camera system equipped to project light beams in order to train on the objects to be captured in poor light conditions or during the night.
It is yet another object of the present invention to provide an intelligent camera system to capture pre-defined universal events, including but not limited to intrusion by animals or humans in restricted areas, patients in distress, malfunctioning of an industrial process, harvest readiness of crops, violation of traffic rules by vehicles, etc.
SUMMARY OF THE INVENTION
The present invention provides a smart camera based system for visual analytics, such as automatic number recognition, visual anomaly, and specificity detection. It comprises of an advanced smart camera unit designed to detect, deter, and disrupt illicit activities or unfortunate events. This system leverages video analytics-based technology deployed with AI and ML embedded into it for automatic and continuous control of the camera’s attributes such as the focal length, aperture, and shutter speed, ensuring optimal image qualities at all times. The system is equipped with external wipers, defoggers, and dehumidifiers to ensure proper camera functioning and optimal video quality in all weather conditions. The system is connected to a cloud server where the live video is processed, and bookmarks are created when a subject or instance is captured by the camera system. In preferred embodiment, the bookmarks are created when a vehicle is captured and then the bookmarks enable the user to view the video clip by selecting the vehicle number within a specific time duration for analysis and dispute resolution at a later point in time. The system is also capable of extracting license numbers from the vehicle’s image and can be trained with several inputs for use in various scenarios.
The present invention enables customizable features that are integrated with hardware components to extend its application. This system has the capability to handle and retrieve text from both non-reflective and reflective number plates, regardless of the lighting conditions. A custom-built OCR is deployed into the current system to recognize different font types and sizes, as well as handwritten number plates. It features multilingual recognition for regional Indian languages, and it can recognize the color of the number plate even during dark lighting conditions and at night. The present invention has a 100% accuracy rate in reading standard number plates as per the laws and regulations of the country. Furthermore, equipped with FOTA capability, this system allows effortless wireless firmware upgrades. It also has substantial processing power that supports video formats of H.265 up to 4K resolutions at 60 FPS, enabling it to record and process high-quality video files to be used at a later point in time during dispute resolution. Data processing in this system occurs swiftly through edge-level computation and recognition. Furthermore, the system is equipped with local logs and edge computation, enabling it to continue functioning even in the event of internet failure or loss of connectivity to the central server.
The present invention is designed to function flawlessly during adverse weather conditions that might interfere with the visual data being captured. For this purpose, the present invention provides a camera system that is equipped with a wiper, defogger, and dehumidifier to keep the lenses of the camera clean. It further has an IP66 rating for its enclosure, which makes it suitable for use outdoors in various weather conditions. In one embodiment, this system also has the ability to control the boom barrier based on the inputs received from the camera sensor and processing the same. The present system features a color camera of 2 ~ 5 Megapixels (“MP”) with an automatically 3.6mm adjustable focal length that supports high-quality video formats. Additionally, the camera lenses provide a wide recognition angle, ensuring that the system captures a wide FOV for accurate and efficient data processing. This system can be mounted on the wall, pole, or even the ceiling providing a versatile solution. The present invention provides an advanced smart camera system that offers a range of features and capabilities that are not available in the existing arts, it is designed to be highly reliable, accurate, and efficient. It can be used in a variety of applications, including but not limited to parking lot management, vehicle inventory management, gas stations, entrance and exit management, and law enforcement.
In the preferred embodiment, the custom-built OCR system can recognize different font types and sizes, as well as handwritten number plates, while its multilingual recognition capabilities make it highly versatile in various non-ideal scenarios. The rugged construction and advanced weatherproofing ensure that the system can operate flawlessly in challenging outdoor environments, making it an excellent addition to any security or surveillance system.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 illustrates a block diagram of the smart camera and edge computing hardware system, depicting the Smart camera (101); Edge computing hardware (102); Communicator (103); and Application on server (104).
Figure 2 illustrates a detailed block diagram of the individual components of the smart camera and edge computing hardware system, depicting the Smart camera (101); Edge computing hardware (102); Enclosure (105); Monitoring and control mechanism (106); Wiper (107); Lens with a shutter (108); Light control unit (109); CMOS/ CCD sensor chip for video/ image capture (110); Controller (111); and Extractor (112).
While the invention is amenable to various modifications and alternative forms, specifics thereof have been shown by way of examples in the drawings and will be described in detail. It should be understood, however, that the intention is not to limit the invention to the particular embodiments described. On the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the invention.
Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. Prior to this, terms and words used in the present specification and claims should not be construed as limited to ordinary or dictionary terms, and the inventor should appropriately interpret the concept of the term appropriately to describe its own invention in the best way. The present invention should be construed as meaning and concept consistent with the technical idea of the present invention based on the principle that it can be defined. Therefore, the embodiments described in this specification and the configurations shown in the drawings are only the most preferred embodiments of the present invention and do not represent all the technical ideas of the present invention. Therefore, it should be understood that equivalents and modifications are possible.
DETAILED DESCRIPTION OF THE INVENTION WITH RESPECT TO THE DRAWINGS
The present invention as embodied by "A smart camera based video analytics system to provide high-quality video and method thereof" succinctly fulfills the above-mentioned need(s) in the art. The present invention has objective(s) arising as a result of the above-mentioned need(s), said objective(s) being enumerated below. In as much as the objective(s) of the present invention are enumerated, it will be obvious to a person skilled in the art that, the enumerated objective(s) are not exhaustive of the present invention in its entirety and are enclosed solely for the purpose of illustration. Further, the present invention encloses within its scope and purview, any structural alternative(s) and/ or any functional equivalent(s) even though, such structural alternative(s) and/ or any functional equivalent(s) are not mentioned explicitly herein or elsewhere, in the present disclosure. The present invention therefore encompasses also, any improvisation(s)/ modification(s) applied to the structural alternative(s)/ functional alternative(s) within its scope and purview. The present invention may be embodied in other specific form(s) without departing from the spirit or essential attributes thereof.
Throughout this specification, the use of the word "comprise" and variations such as "comprises" and "comprising" may imply the inclusion of an element or elements not specifically recited.

KEY DEFINITIONS:
• ANPR: Automatic Number Plate Recognition
• OCR: Optical Character Recognition
• IoT Server: Internet of Things Server
• CMOS: Complementary Metal-Oxide-Semiconductor
• CCD: Charge-Coupled Device
• AI: Artificial Intelligence
• ML: Machine Learning
• GPS: Global Positioning System
• IR Filter: Infrared Cut-off Filter
• Wi-Fi: Wireless Fidelity
• GSM: Global System for Mobile Communications
• GPRS: General Packet Radio Service
The present invention provides a smart camera and edge computing hardware system to provide high-quality video for computer vision. The present system not only captures images and videos but also identifies anomalies by comparing them with existing images or videos that are stored in the database. The present invention provides an optimal intelligent visual capturing module that incorporates automatic positioning and data training on the object of interest and optimizes itself for obtaining crisp images of the images that are likely to be blurred due to sub-optimal environmental conditions.
The deep image processing technology of the present system improves the images’ qualities significantly, the various steps in the image processing include:
• Visualization – Finding objects that are not visible in the image.
• Recognition – Distinguishing or detecting objects in the image.
• Sharpening and Restoration – Creating an enhanced image from the original image.
• Pattern Recognition – Measuring the various patterns around the objects in the image.
• Retrieval – Browsing and searching for images from a large database of digital images that resemble the original image obtained during capture.
There are various practical challenges that arise while carrying out the above-mentioned process, for example, the environmental conditions during capturing the image or the video might hinder the capture quality (e.g., unfavorable lighting conditions, humidity, rainfall) or any external factors tend to affect the image quality making it difficult to be processed in the due course. Additionally, there is a high bandwidth requirement for carrying out the above processes, and then arises a need for a large storage to process high volumes of data and compute. Furthermore, to process the images there is a need for complex hardware requirements that increase the overall cost of the system.
Provided below is the list of components of the present invention along with its reference numerals:
COMPONENT REFERENCE NUMERAL
Smart Camera Unit 101
Edge Computing Hardware 102
Communicator 103
Application on Server 104
Enclosure 105
Monitoring and Control Unit 106
Wiper 107
Lens with Shutter 108
Light Control Unit 109
CMOS/ CCD Sensor Chip for Video/ Image Capture 110
Controller 111
Extractor 112
The system disclosed by the present invention comprises of a smart camera unit (101) and an edge computing hardware (102) working together with a communicator (103) which is communicably coupled with an application running on a server (104) with an in-built database that stores the information, to realize the entire functionality of various embodiments where the smart camera (101) and edge computing hardware (102) are deployed.

FLOW OF ACTIONS CARRIED OUT BY EACH COMPONENT WITH REFERENCE TO FIGURE 1
Smart Camera (101):
According to the preferred embodiment, the smart camera (101) is at the head of the chain of actions that are performed, and the various steps that are involved in the smart camera (101) based video analytics system are as follows:
1. Tracking – Finding the object of interest that are not directly positioned in the frame of the camera’s vision.
2. Locating & Recognizing – Detection and confirmation of the object of interest within the frame of the camera vision.
3. Extraction – Extracting the image from the video for subsequent processing and analysis.
4. Image Quality Monitoring – Evaluating the quality of the image that has been captured for further enhancement.
5. Image Quality Enhancement – The quality of the image is enhanced by automatically/ manually controlling various components and attributes of the smart camera (1) such as, the wiper, defogger, dehumidifier, focal length, aperture shutter speed, etc.
6. Repetition of steps 3 – 5 until obtaining the image quality above a threshold value.
Edge Computing Hardware (102):
Subsequent processing is performed at the edge computing hardware (102) of the smart camera (101) system following a series of steps as described below:
Firstly, the quality of the image that has been captured is assessed. Upon assessment, if the image is not of optimal and required quality, instructions are sent back to the smart camera (101) to obtain an image of enhanced quality by controlling the components and attributes of the smart camera (101).
Secondly, the patterns of contours of the object in the frame match statistically with the contours of the standard objects that are stored in the database of the machine learning module.
Thirdly, if the matched object of interest is within a statistical distance of the stored standard object, it is statistically updated, and the extracted object contour is further processed.
Fourthly, if any text must be identified in the application, the image that contains the text is extracted and converted into a digital format which is further processed using the tailored OCR alongside AI/ ML models. Furthermore, if a human face has to be extracted, or an industrial object has to be extracted, such trained models with specific instructions are deployed into the system which can identify and extract the same. The object of interest can vary and be vast as per the use case where the smart camera (101) system is deployed.
Finally, the extracted information, pictorial or text, a single image, or multiple images forming a video clip is further transferred onto the communicator (103) to pass on to the server over the network.
Communicator (103):
The communicator (103) hardware, upon receiving the extracted information of text data or picture data from the edge computing hardware (102), further converts it into an encoded stream of data and connects to the server to upload the encrypted stream to the application on server (104).
Application on server (104):
The application on server (104) deployed on the cloud processes the input data that is received for subsequent use. For instance, in the preferred embodiment of the present invention during an ANPR application, the server receives the text data, which in this use case is the number of a vehicle whose image depicts the vehicle violating the rule by using the wrong lane, the application on server (104) processes it for finding the owner details of the vehicle and sending a violation ticket demanding for payment of the fine.
In another embodiment of the present invention, if the received data is the image of a patient in distress, which is identified by their facial expression with trained data, the application on the server (104) alerts the appropriate healthcare official for immediate medical action.
In another embodiment of the present invention, if the image depicts an intruder in a restricted area, the application on the server (104) alerts the security official. The use case of the present invention is vast according to the input and training models.
The smart camera unit (101) is equipped with sensors and components to control various parameters of the camera system to obtain the best possible image output quality by altering the camera’s attributes such as the focal length, shutter speed, etc. alongside various adjustments made to the image quality by employing the wiper, defogger, and dehumidifier during sub-optimal environmental conditions. As depicted in Figure 1, the smart camera (101) captures the image/ video of the object of interest and extracts the image from the environment where it is deployed.
It is pertinent to note that, due to various reasons, including non-optimal weather conditions such as rainfall, fog, poor lighting, glare, dust, and smoke, etc., or because of the misconfigured camera settings, such as the area on which the camera is trained or focus points, shutter speed, brightness, contrast, etc. the quality of the captured image is likely to be poor for any information of interest to be extracted from it. There might be circumstances where the object of interest may not be visible in the camera’s FOV at all or might move out of the camera’s FOV. The present invention solves all the aforementioned drawbacks with the use of smart camera components and controlling the camera's attributes.

DESCRIPTION OF EACH COMPONENT OF SMART CAMERA UNIT (101) AND THE EDGE COMPUTING HARDWARE (102) WITH REFERENCE TO FIGURE 2
Smart Camera unit (101) comprises the following components embedded in it:
1. Enclosure (105)
The enclosure (105) including its mounting assembly encloses the components of the smart camera unit (101). It protects the camera from outside environmental disturbances and a transparent screen at the front seals the camera making it dust-proof and water-resistant with an IP66 rating.
2. Monitoring and control unit (106)
This unit comprises the sensors and actuators to monitor various conditions and change the relative configuration of the components or change the settings of the camera to enhance the quality of the image. This decision-making based on parameters that are monitored, and the control action commands that are to be generated are processed by the controller (107) in the edge computing hardware (102).
3. Wiper (107)
The wiper (107) is mounted on the transparent screen of the camera system and is driven by a micro-motor that in turn is controlled by the signals from the edge computing hardware (102). The wiper is toggled to remove any water, fog, snow, or other debris from the camera lens thereby ensuring clear images in non-optimal weather conditions.
4. Lens with a shutter (108)
The lens with a shutter (108) of the camera is mounted on rails and is designed to be moved back and forth by a micro-motor that is controlled by the controller (111) in the edge computing hardware (102).

5. Light control unit (109)
The light control unit (109) is designed to detect poor lighting conditions which might be due to adverse weather conditions during daytime or low lighting conditions during the night. Subsequently, the light control unit (109) turns on the auxiliary external lights of the camera system. Similarly, the bright sunlight glare is also detected by the sensors and the IR filter control is carried out in order to reduce the brightness of the scene that is being captured.
6. CMOS/ CCD sensor chip for video/ image capture (110)
The CMOS/ CCD sensors (110) are the core element of the camera system that captures the video or the image when exposed. These sensors convert the photos striking the sensor array into digital signals that are further processed to recreate a digital representation of the scenario that is captured in the video or photo mode. The captured video is subsequently transferred to the controller (111) of the edge computing hardware (102) where the quality of the image/ video is verified, and the control signals are transmitted back to the smart camera (101) components to enhance the quality of the image.
Edge computing hardware (102) comprises the following components embedded in it:
1. Controller (111)
The controller (111) functions towards obtaining the optimal quality of the image/ video that is possible. Such high-quality images are of utmost importance for the extractor to be able to extract relevant data/ information from the footage. The controller (111) is a microprocessor-based electronic unit that receives the footage from the smart camera (101) and evaluates its quality. If the image obtained does not meet the required quality, it verifies the information received from the sensors and transmits signals to activate action to alter the camera’s attributes or trigger its climate control components to enhance the quality.
2. Extractor (112)
The extractor (112) module further extracts the image frames from the video, zeroes in on the object of interest, i.e., a vehicle, human, industrial machine, or animal, etc., and extracts the image of the object. Depending on the use case, it further locates the parts of the image, for example, for monitoring vehicles, it extracts the number plate and image of the number. When deployed for surveillance, it locates the face of the human being and further extracts the facial features to detect and deter fraudulent activities. When deployed at industrial plants, it will further zero in on a sub-part of the equipment that is suspected to be malfunctioning, etc. By this hierarchical processing of sub-parts of the targeted object, the extractor (112) extracts the data/ information that is required.
DESCRIPTION OF INTERWORKING OF THE COMPONENTS
The controller (111) in the edge computing hardware (102) and the components in the smart camera unit (101) work together in order to achieve the best quality data, described below is the functionality of the present invention that overcomes the drawbacks of the prior art and further elucidates on how the extractor (112) extracts data/ information.
The controller (111) addresses two sets of challenges, the first challenge pertains to the
non-optimal environmental conditions that are present which needs to be addressed to prevent poor quality of image/ video capture. The other challenge pertains to enhancing the quality of the captured footage by automatically adjusting the attributes and settings of the camera such as the focus, brightness, shutter speed, aperture, etc.
The sensors in the smart camera (101) detect conditions such as rainfall, fog, low light, formation of cobwebs, mist, dust, etc., on the front protective transparent screen. When these conditions are detected by the sensor units, the signals are transmitted to the controller that sets in motion control actions to correct the disruptions to the view by toggling the climate control components embedded within the camera, such as the wipers, defoggers, dehumidifiers, etc. For instance, when such disruptions are detected on the front transparent screen, the micro-motor that controls the wiper is turned on consequently toggling the wiper to perform its function. In another instance, if the light control unit (109) detects inadequate lighting or glare due to a highly intense light source, such as the sunlight, it transmits signals to the controller (111) to activate the IR filters thereby reducing the effect of the glare. If the sensor units detect fogging on the transparent screen, the defogger or the heating element is turned on in the camera system accordingly through the instructions transmitted by the controller (111). Furthermore, the challenge could also be due to the object of interest not falling in the camera’s frame/ FOV, in such circumstances, the controller (111) transmits instructions to move the entire enclosure in an angular trajectory and also swivels the camera up and down by controlling the motor in the enclosure mounting assembly thereby pointing the camera towards the object of interest at the center. These steps are repeated until the optimal image angle is obtained and the object of interest falls within the camera’s FOV.
More often, the image/ video quality can be enhanced by adjusting the camera’s settings/ attributes, such as the focal length, shutter speed, aperture, exposure time, brightness, contrast, etc. In order to facilitate this, the camera lenses are moved by a micro-motor driven by the controller (111) to adjust the distance between the camera lens and the CMOS/ CCD sensor chip for video/ image capture (110) in order to obtain the footage with perfect focus to the object of interest. Additionally, if the video footage is over-exposed or under-exposed, the controller adjusts the shutter speed and/ or the aperture, with the help of the micro-motor which is activated and driven by the controller (111) until the desired aperture is obtained.
The present smart camera system is connected to the application on the server (104) both via Wi-Fi Network and/ or GSM Network. The present invention has a trained AI/ ML model incorporated within it alongside an intelligent remote condition optimizer. The smart camera (101) is connected to the application on server (102) over local/ Wi-Fi/ GSM networks which enables image processing where the backend decisions and processing are made, and instructions are fed back to toggle the climate control components and alter the exposure triangle including hue, saturation, and other image parameters such as adequate lighting, proper camera calibration, and efficient image pre-processing, for obtaining optimal image/ video footage. The present system has an intelligent hardware enclosure that can optimize and modify the camera’s attributes remotely from the backend while viewing and comparing the existing environmental conditions at the time of capture and in such wise, the system aids in optimized image capture which can improve several applications.
In the preferred embodiment of the present invention, the system is trained with an ML model for the recognition of number plates for ANPR application, and it can easily recognize various complex linguistics and a multitude of number patterns that are available in vogue. The present system with a well-trained ML model that can detect vehicles can also recognize the placement of the number plate, as it is trained with various sample data of the vehicles and the number plate’s position being identified during training.
The extractor (112) extracts the font/ text from the number plate and also facilitates the controller (111) to work with the backend server to ensure that the optimizers and climate control components can get the right data and further ensure that the parameters are optimized in real time. The values of the settings are corrected and consequently, the high-quality image that is obtained is transmitted back to the server for further processing.
As soon as the system detects a moving vehicle, it captures the image with optimal parameters and extracts the text from the license plate. The segmented characters are passed through an OCR engine that is embedded within the ML model for further identification. Furthermore, the shape and pattern of the characters are ascertained to provide accurate predictions. In this use case, the AI/ ML models within the system have specific models developed for various countries and regions. For instance, in India, multilingual multi-fonts and number plates are placed in multiple locations, and the training data is provided in order to facilitate detection and extraction from such number plates.
In the preferred embodiment of the present invention, the smart camera (101) is triggered to capture images when a vehicle enters a predefined detection zone, and this is achieved through a motion detection sensor embedded into the hardware of the embodiment. In action, when a vehicle triggers the camera through its motion, the camera captures a series of images in rapid succession/burst mode. These images are then used for processing stages as stated above. The multiple images are obtained to increase the likelihood of capturing clear and usable images, especially when the vehicle is moving at high speeds. The camera unit in this present system can accurately capture the license plates even while the vehicles are moving at high speeds.
Conventionally, the camera’s accuracy depends on the lighting conditions, weather, the quality of the camera and lens, and the efficiency of the OCR and processing models. Adequate lighting, proper camera calibration, and efficient image pre-processing play a crucial role in achieving accurate recognition results. In order to achieve this, the present invention takes into account all the existing parameters and optimizes the parameters in
real-time such that enhanced and efficient images are captured. This is carried out by the AI/ ML model that gives the inputs to the controller (111) which controls the wiper, defogger, dehumidifier, light sensors, heater and camera’s attributes in order to ensure that the images that are captured are of the best quality for further processing. Once the images are obtained, it is pre-processed by the edge computing hardware (102) and then processed through AI/ ML models to determine if the object of interest has entered the zone, upon verification and identifying the other data from the image such as, the number plate’s position in the vehicle, the texts from the number plate are captured and forwarded to the OCR engine in ANPR applications. The OCR engine thereby identifies the characteristics, extracts the text, and forwards it to the system. Subsequently, the data is sent to the central server on the cloud along with the GPS coordinates via Wi-Fi or GPRS for further processing. The application deployed on the server (104) has an inbuilt workflow for processing.
The present invention aids in improving the environmental conditions while capturing, such as the lighting, and humidity, at the hardware level itself. These features are inbuilt into the hardware structure that has been explained above. The present invention is a fixed system that can be mounted on walls, poles, etc., adheres to multiple fast-moving vehicles for ANPR applications, monitoring patients, monitoring industrial plants, etc., and improves the visual parameters remotely to optimize the images captured dynamically. It further provides a comprehensive solution to address various challenges described above and improves the accuracy of the system by providing means for automated real-time adjustments of the visual parameters of the surrounding conditions. As the embodiment is designed in-house it proves to be cost-effective.
EXAMPLE
The present invention discloses a smart camera with an edge computing hardware system and a method to provide a video analytics-based solution that is intended to detect visual anomalies and also deter illicit activities by implementing artificial intelligence and machine learning. The system comprises of smart camera (101) coupled with the edge computing hardware (102), along with the communicator (103), that is communicably coupled with an application on server (104). The edge computing hardware (102) further comprises of sub-components: the enclosure (105) that is IP66 rated protecting the smart camera and connected components from outside environmental disruption; monitoring and control unit (106) that comprises of the sensors and activators to monitor the environmental conditions and change the relative configuration of the camera; a wiper (107) connected to a micro-motor for wiping the transparent screen covering the camera system; a lens with a shutter (108) connected to a micro-motor to control the shutter speed, light control unit (109) to detect poor/ bright lighting conditions and toggle the auxiliary external lights or IR filter of the camera; a CMOS/CCD sensor chip for video/ image capture (110); a controller (111) comprising a microprocessor based electronic unit to receive inputs from the camera system and evaluate the quality and send instructions back to the actuators for corrective actions to enhance the image; and the extractor (112) for further extracting the image and zeroing it down to the object of interest. Furthermore, the method of visual analytics comprises the following steps:
(i) Firstly, the smart camera (101) detects the object of interest that may or may not be directly in the centre frame of the smart camera’s (101) vision. Upon detection, the object of interest is located and recognized, and the wide FOV of the camera unit facilitates this process. Subsequently, the camera adjusts its positioning, focus, and along with other calibrations to zero down on the object of interest.
(ii) Secondly, the image/ video is extracted to be processed by the monitoring and control unit (106) that evaluates the quality of the image that is captured.
(iii) subsequently, the controller (111) controls various components of the camera system such as the wiper (106) defogger, dehumidifier, heater, light control unit (109), etc., to enhance the image quality.
(iv) The above-mentioned extraction, monitoring, and enhancement steps are repeated until the image/ video quality is above a predefined threshold value.
(v) Concurrently, the edge computing hardware (102) assesses the output image quality, and if the quality of the image is found to be inadequate instructions are sent to the controller (111) which subsequently triggers the camera unit to provide an image of enhanced quality. Further, the patterns of the contours of the object in the frame are statistically matched with the contours of the standard objects that are stored in the trained ML database.
(vi) The object of interest along with the text (if any) are identified and if the text is available, it is further extracted by the extractor (112) and converted into a digital format which is further processed by the OCR engine.
(vii) Finally, the extracted information is then passed on to the communicator (103) and for further transmission to the application on the server (104). The communicator hardware receives the extracted information of text or picture from the edge computing hardware (102), converts it into an encoded stream of data, connects to the application on server (104), and uploads the encrypted stream to the server (104) on cloud that processes the data that is received for subsequent use.
In the preferred embodiment, the smart camera (110) is triggered to capture images when a vehicle enters a predefined detection zone, and this is achieved through motion detection embedded into the hardware of the embodiment. In action, when a vehicle triggers the camera, the camera captures a series of images in rapid succession/ burst mode. These images are then used for processing stages as stated above. The multiple images are obtained to increase the likelihood of capturing clear and usable images, especially when the vehicle is moving at high speeds.
The obtained images are pre-processed by the server which further determines if the image is of a vehicle that has entered the zone. Upon verifying and identifying the number plate’s position in the vehicle, the texts from the number plate are captured and forwarded to the OCR engine. The OCR engine thereby identifies the characteristics, extracts the text, and forwards it to the system. Once the vehicle registration number is determined, the data is sent to the central server on the cloud along with the GPS coordinates via Wi-Fi or GPRS for further processing. The application deployed on the server (104) has an inbuilt workflow for processing. The example above demonstrates the use case of the present invention, which can be trained on multiple objects of interest as needed and tailored accordingly to meet various requirements.
ADVANTAGES OF THE PRESENT INVENTION
• The system improves the environmental conditions while capturing, such as the lighting, and humidity, at the hardware level itself.
• This is a fixed system that can be mounted on walls, poles, etc., and adheres to multiple fast-moving vehicles, patient monitoring, industrial plants monitoring, etc., and improves the visual parameters remotely to optimize the images captured dynamically.
• Provides automated real-time adjustments of the visual parameters of the surrounding conditions.
• The present invention’s embodiment is designed in-house on the existing IoT platform, and hence the present invention also proves to be cost-effective.
• The present visual analytics system focuses on improving and optimizing the parameters of the image captured, remotely.
• It identifies visual anomaly and specificity detection by comparing the images from the existing database and provides optimal identification of the images that are captured.
• The present system is dynamic, it not only ensures improved accuracy, but it also provides periodic monitoring features to ensure that the images that are captured are of the highest quality for the detection of anomaly.
• The present invention also maintains an audit trail of all the events of images.
• The rugged construction and advanced weatherproofing of the system ensure that the system can operate flawlessly in challenging outdoor environments, making it an excellent addition to any security or surveillance system.
In most of the use cases, when cameras are deployed remotely, the live videos are monitored in a centralized observation and managing facilities that require human supervision. The present invention leverages computer vision technology and facilitates automating the monitoring mechanism by providing information/ alerts extracted from the images/ video footage. For example, when deployed on highways for monitoring vehicles and traffic violations, the present invention can return the vehicle license number extracted from its number plate when there is over-speeding, lane violation, or other similar scenarios which can be tailored according to the laws and needs of the region. Similarly, the present invention can detect and identify an intruder by extracting the face of the intruder and comparing it with a trained database. As the cameras are many a time mounted in remote and inaccessible places, the efficiency of the entire computer vision system is likely to be affected due to the poor quality of image/ video that is captured. In such, cases if such
low-quality videos are monitored and supervised by humans in centralized observation facilities, the authorities may not be able to control or give commands for corrective action in a short time period, which is very crucial to detect and deter anomalies. Therefore, the present invention provides an advanced smart camera with edge computing hardware to counter this problem by ensuring that the quality of the footage is optimal even under challenging conditions.
Although the proposed concept has been described as a way of example with reference to various models, it is not limited to the disclosed embodiment and that alternative designs could be constructed without deviating from the scope of invention as defined above.
It will be apparent to a person skilled in the art that the above description is for illustrative purposes only and should not be considered as limiting. Various modifications, additions, alterations, and improvements without deviating from the scope of the invention may be made by a person skilled in the art.
, Claims:We Claim,
1. A smart camera based video analytics system to provide high-quality video for computer vision, characterized in that:
a. A smart camera unit (101) provided with sensor chips (110) that capture video or image of static or moving target, comprising of:
An enclosure (105) provided with a mounting assembly, and further encloses the camera with a transparent screen on the front side of the camera;
A monitoring and control unit (106) comprising of sensors and actuators that monitors the camera unit (101) to enhance the quality of the image;
A lens with a shutter (108) mounted on rails and moved back and forth by a micro-motor;
A light control unit (109) that detects lighting conditions of the environment;
b. An edge computing hardware (102) provided with a controller (111) for obtaining the optimal quality of the image/ video and further provided with an extractor (112) module to extract the image frames from the captured video;
c. A communicator (103), wherein the communicator (103) upon receiving the extracted information of text data or picture data from the edge computing hardware (102), further converts it into an encoded stream of data and communicates the said data to a server (104);
d. A server (104) which is communicably coupled to the communicator (103), is provided with an application that processes the data stored in a database.

2. The smart camera system as claimed in claim 1, wherein the camera unit (101) is provided with a wiper (107) on the transparent screen of the enclosure (105) and is driven by the micro-motor that in turn is controlled by the signals from the edge computing hardware (102) through the controller (111).

3. The smart camera system as claimed in claim 1, wherein when disruptions are detected on the front transparent screen of the camera unit (101), the micro-motor driven by the controller (111), controls the wiper (107) and is turned on consequently toggling the wiper to perform its function.

4. The smart camera system as claimed in claim 1, wherein when the light control unit (109) through sensors detects inadequate lighting or glare due to a highly intense light source such as the sunlight, it transmits signals to the controller (111) to activate IR filters thereby reducing the effect of the glare on the camera unit (101).

5. The smart camera system as claimed in claim 1, wherein when the sensor units (110) detect fogging on the transparent screen of the enclosure (105), a defogger or a heating element is turned on in the camera unit through the instructions transmitted by the controller (111).

6. The smart camera system as claimed in claim 1, wherein the controller (111) is a microprocessor based electronic unit that receives inputs from the camera unit (101) and evaluates the quality of the images and send instructions back to the actuators for corrective actions to enhance the parameters of the captured image remotely and in real-time.

7. The smart camera system as claimed in claim 1, wherein the micro-motor is driven by the controller (111) to obtain the desired aperture of the lens (108).

8. The smart camera system as claimed in claim 1, wherein the camera unit (101) is provided with a motion detection sensor which is triggered to capture images when a vehicle enters a predefined detection zone.

9. The smart camera system as claimed in claim 1, wherein the system is connected to the application on the server (104) via Wi-Fi Network and/ or GSM Network.

10. The smart camera system as claimed in claim 1, wherein the parameters of the image/video include hue, saturation, adequate lighting, proper camera calibration, image pre-processing.

11. The method of working of the smart camera based video analytics system to provide high-quality video, comprises of:
a. The smart camera (101) detects the object of interest that may or may not be directly within the centre frame of the smart camera’s (101) vision;
b. Subsequently, the camera adjusts its positioning and focus to zero down on the object of interest through its wide field of view (FOV);
c. The image/ video is extracted and processed by the monitoring and control unit (106) that evaluates the quality of the image that is captured;
d. Subsequently, the controller (111) controls various components of the camera system such as the wiper (106) defogger, dehumidifier, heater, light control unit (109) to enhance the image quality;
e. The above steps of extraction, monitoring, and enhancement are repeated until the image/ video quality is above a predefined threshold value;
f. the edge computing hardware (102) assesses the output image quality, and if the quality of the image is found to be inadequate, the instructions are sent to the controller (111) which subsequently triggers the camera unit (101) to provide an image of enhanced quality;
g. the patterns of the contours of the object in the frame are statistically matched with the contours of the standard objects that are stored in the trained database on the server (104);
h. object of interest along with the text (if any) are identified and if the text is available, it is further extracted by the extractor (112) and converted into a digital format which is processed by an OCR engine;
i. The OCR engine thereby identifies the characteristics of the image, and extracts the text;
j. The extracted information is then passed on to the communicator (103) for further transmission to the application on the server (104);
k. The communicator (103) receives the extracted information of text or picture from the edge computing hardware (102), converts it into an encoded stream of data, communicates to the application on the server (104), and uploads the encrypted data to the server (104) on cloud along with GPS coordinates, that processes the received data for subsequent use.

12. The method as claimed in claim 11, wherein when a vehicle triggers the camera (101) through the motion detection sensor, the camera (101) captures a series of images in rapid succession which are used for processing.

13. The method as claimed in claim 11, wherein bookmarks are created when a vehicle’s video is captured, wherein the bookmarks enable a user to retrieve and view the video by selecting the vehicle number within a predefined time duration for analysis.

14. The method as claimed in claim 11, wherein the sensors (110) convert the images striking the sensor array into digital signals that are further processed to recreate a digital representation of the image that is captured in the video or photo mode, which is subsequently transferred to the controller (111) where the quality of the image/ video is verified, and the control signals are transmitted back to the smart camera unit (101) to enhance the quality of the image.

15. The method as claimed in claim 11, wherein, to position the object of interest within the camera’s FOV, the controller (111) transmits instructions to the camera unit (101) to move the entire enclosure assembly (105) in an angular trajectory and also swivels the camera unit (101) up and down by controlling the motor in the enclosure (105) thereby pointing the camera towards the object of interest at the centre.