Method for detecting, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, estimating weight, grading and mapping of objects

Abstract

The present invention provides a computer-implemented method for detecting, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, estimating weight, grading and mapping of at least one predetermined object in real-time mode, the method comprising obtaining or capturing a plurality of visual data from a surrounding area using at least one imaging device (101), transmitting the plurality of visual data to an at least one processing unit (104) of a computing device, identifying and detecting the at least one predetermined object trained with at least one AI machine learning algorithm using the plurality of visual data, grabbing and picking of the at least one predetermined object by the at least one grabbing means (105), obtaining at least one geolocation of the at least one predetermined object by way of at least one navigation system (106), counting the at least one predetermined and annotating, classifying, estimating weight and grading of the at least one predetermined object.

Description

[0001] METHOD FOR DETECTING, GRABBING AND PICKING UP, COUNTING, OBTAINING A GEOLOCATION, ANNOTATING, CLASSIFYING, ESTIMATING WEIGHT, GRADING AND MAPPING OF OBJECTS FIELD OF INVENTION The present invention generally relates to detecting, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, estimating weight, grading and mapping a predetermined object. More specifically, the present invention relates to a method and system for detecting, grabbing andpicking up, counting, obtaining a geolocation, annotating, classifying, estimating weight, grading andmapping of oil palm fresh fruit bunches based on a first classification, a second classification and a third classification. BACKGROUND OF INVENTION Harvesting oil palm fruits in the estate involves several steps. Firstly, the oil palm fronds are cut, followed by the cutting of the ripe oil palm fresh fruit bunches (FFBs). Then, the cut oil palm fronds are stacked around the oil palm trees, while the cut oil palm FFBs are arranged along the harvesting path for in-field collection and evacuation. A movable apparatus, vehicle, or prime mover, such as a mini tractor with a grabber, is used to transport the collected oil palm FFBs to a designated collection bin in the field. Lastly, a hook-lift lorry is employed to evacuate the collection bin and transport it to a palm oil mill for further processing. At the mill ramp of a palm oil mill, before processing the oil palm FFBs, a trained human grader conducts random checks to assess the ripeness level and detect any rodent damage in the batches of FFBs brought in for processing. The grader maintains a record of the condition and location of the FFBs, including which oil palm estate they were collected from. Based on this information, the estate management can take necessary actions if the FFBs fail to meet the required standards. At an oil palm mill, the conventional method of grading the oil palm fresh fruit bunches (FFBs) involves a skilled worker stationed at the mill ramp. This worker observes, checks and grades the FFBs depending on their ripeness level and any damage caused by rodents. The conventional means of checking and grading oil palm FFBs has several disadvantages. For instance, it only provides a rough estimate for a large area of an oil palm estate as not all the FFBs are checked and graded. Usually, only about 20% of a load is checked at the mill ramp. This is because it would be impossible for all FFBs to be graded manually without automation. For instance, a 60T / h mill can process around 1,000T FFB per day. It would be humanly impossible to grade all of these at a 100% rate. Tracing the original location of oil palm FFBs is not possible through conventional means. Moreover, the consistency and accuracy of the grading process varies between workers, as each grader can have different experiences, leniency, attitude and perspectives when grading the FFBs at the mill ramp. The data is manually recorded in logbooks, which require storage space and are prone to errors. The integrity of the grader plays a significant role in the accuracy of the recorded data. Additionally, the conventional means of grading FFBs is not efficient, optimized, or accurate in determining the ripeness levels and rodent damage of the oil palm FFBs. A computer-implemented method for improving crop harvesting is described in PCT Publication WO2021 / 225528. The method involves collecting data from one or more sources, extracting features from the data, determining a quality grading based on the extracted features, generating a harvestingdata model using the collected data, extracted features, and the determined quality grading, and makingoperational decisions using the model. This method can be implemented on a system with at least one memory device and processor or a computer-readable storage medium that contains instructions to be executed with at least one memory device and processor. The United States Patent Application US20220172466 explains a control system designed for precision agriculture. The system includes a controller equipped with a convolutional neural network, an imaging device, a variety of sensors, and multiple actuators. The controller is configured to detect and track objects of interest in a crop field, such as crops themselves. The system's multiple agricultural tool actuators are operated based on the detected and tracked objects of interest in the crop field. An Indian patent application, identified as IN202241015310, outlines an innovative approach to satellite technology that is able to capture high quality field images for precision agriculture. The invention is based on the use of Artificial Intelligence (AI), specifically machine learning (ML) and deep learning, to automatically predict crop yield through image processing and other AI methods. The invention comprises three modules: an image capture module, a data analytics module, and a visualization module. The image capture module uses satellites launched by ISRO to capture high quality graphical imagery of specific crops, while the data analytics module employs a Convolutional Neural Network (CNN) based deep pipeline to process the data and estimate crop yield. Finally, the visualization module presents the outcomes of the invention. The deep learning model uses pre-trained models with transfer learning to achieve better prediction of crop yield. This invention has the potential to benefit numerous stakeholders, including farmers, agricultural departments, governments, agricultural universities, researchers, and academic institutions. The United States Patent US8381501 explains how an agricultural robot system can help with harvesting, pruning, culling, weeding, measuring, and managing of crops. Robots that are autonomous or semi-autonomous, use machine vision through cameras to identify and locate fruit, or points on the vine that need pruning. The cameras can also be used to measure different agricultural parameters or assist in the management of resources. Additionally, the robots can be equipped with an arm or other tools to perform the desired agricultural function. First, a robot surveys the field and maps out plant locations, counts the number of fruit, and determines their size and approximate position. Then, a robot or server can create an action plan specifying the agricultural function to perform based on the gathered data. The action plan may include operations and data required for performing the function. A system for providing suggestions to users is described in Indian Patent Application IN202111054299. The system consists of a data retrieval unit that retrieves a first data set containing information on moisture sensors, climate sensors, and sensing arrangements. A machine learning model is trained using pre-stored data, and the first data set is then trained and tested using the model. Based on the training and testing, a report is generated and sent to a computing device via a communication network.The United States Application US20210235641 presents a system for monitoring and regulating plantproductivity. The system includes a memory for storing instructions and a processor for executing these instructions. The method involves receiving field data from monitoring sensors, computing a predicted value for a variable associated with the production environment condition of a crop field using a machine learning algorithm, and determining, based on a threshold associated with the variable, that an intervention in the crop field is necessary. This intervention is initiated by causing a controllable device to vary the production environment condition. The machine learning algorithm has been trained based on a training set comprising the field data from the monitoring sensors and / or a generated feature derived from the field data. The ripeness grading of oil palm fruit bunches is the subject of a research paper titled "Oil Palm Fruit Bunch Grading System Using Red, Green and Blue Digital Number." Currently, the grading of oil palm bunches is done manually by human graders, which is not only subjective but also prone to disputes. To address this issue, we developed an automated grading system using the RGB color model. The system distinguishes between the three different categories of oil palm fruit bunches based on their color ripening index, which is determined by different color intensities. Our grading system uses a computer and camera to analyze and interpret images just like the human eye and brain. We investigated the colors of Red, Green, and Blue (RGB) of the palm oil fruit bunch and developed a computer program that uses the mean color intensity to differentiate between the different colors and ripeness of the fruit, such as oil palm FFB. The program results showed that the ripeness of the fruit bunch could be differentiated between different categories of fruit bunches based on RGB intensity. [Source: August 2008Journal of Applied Sciences 8(8), DOI:10.3923 / jas.2008.1444.1452] A Malaysian patent, identified as MY-157647-A, outlines a fruit ripeness grading system that uses a computer vision application to inspect the agricultural quality of fruits and ensure their ripeness category. The system comprises a housing with an enclosure for scanning, an illumination system with an optical lens illumination filter located at the enclosure of the housing, a camera located at the top portion of the housing enclosure, a feeding device that conveys fruit samples to the housing, a conveyor speed inverter, a processing unit that analyzes the fruit sample image, and a data acquisition interface located between the camera and the processing unit. The feeding device is further equipped with a separator controlled by a USB controller and supported by a compressor, which is used to control and separate the fruit samples. The processing unit is also equipped with a disk top computational unit that can transfer data to a computer. The fruit ripeness grading system is specifically designed for oil palm fresh fruit bunches (FFB) and uses image processing and analysis to grade and classify fruit samples based on colour, texture, empty sockets, and thorns features. This system provides a non-destructive measurement method that does not require laboratory examination to determine fruit ripeness. It is also portable and can be adopted into industrial chain frameworks. The system enables the classification of large numbers or quantities of fruits with high speed, accuracy, and time-saving benefits. It is important to note that this publication does not describe the process of measuring the ripeness of oil palm fruitlets through real-time chlorophyll content measurement of a plurality of samples. A spectrometer isconfigured to analyze, measure, and convert light signals, including collected fluorescence, reflectance,transmittance, and / or emission of light from excited chlorophyll molecules in a range of samples (such as oil palm fruitlets, mesocarp, mass passing digester, undiluted crude palm oil, diluted crude palm oil, or crude palm oil) into electrical signals that provide the chlorophyll content reading of the oil palm fruitlets. Malaysian Patent MY-162606-A describes an automated system for grading oil palm fruits. The system sorts the fruits into different grades based on their ripeness, using red light laser scanning. The process begins by cutting the fruit skin, after which the fruits are placed on a moving conveyor belt to be transported and graded. As the fruits pass through the laser light, the amount of reflected laser light is measured and converted into digital values. These values are then correlated with the ripeness of the fruit. The patent specifies that the average incident of light levels for under ripe fruit is between 3,500- 4,000, for ripe fruit it is between 3,000-3,500, and for overripe fruit it is between 2,500-3,000. However, the patent does not mention the use of a spectrometer to measure the real-time chlorophyll content of the oil palm fruitlets, which can provide additional information about the fruit's ripeness. Malaysian Patent MY-164318-A describes a system that uses hyperspectral imaging technology to grade and determine the quality of oil palm fruit. The system comprises a housing with an enclosure for scanning, an illumination means for shadow-free illumination, a spectral camera with a CCD array for capturing fruit sample images, a conveying means for providing a scanning platform, a processing unit for analyzing the images, and a data acquisition interface between the camera and the processing unit. The system is characterized by the use of hyperspectral imaging technology and an Artificial Neural Network (ANN) technique for ripeness classification of oil palm FFB. The ripeness classification is done by analyzing fruit samples from three ripeness categories: under ripe, ripe, and over ripe. The enclosure of the housing is made of dark finishing material, and the conveying means is furnished with non- reflective finishing to carry fruit samples into the illumination field of the system. The processing unit uses data processing software such as MATLAB® to perform the analysis of fruit sample classification and obtain resultant quality data. However, the system does not describe measuring the ripeness of oil palm fruitlets via real-time chlorophyll content measurement of a plurality of samples. This involves using a spectrometer to analyze, measure, and convert light signals that comprise collected fluorescence, reflectance of light, transmittance of light, and / or emission of light from excited chlorophyll molecules in the plurality of samples, such as the oil palm fruitlets, mesocarp, mass passing digester, undiluted crude palm oil, diluted crude palm oil, or crude palm oil, into electrical signals to provide readings on the chlorophyll content of the oil palm fruitlets. The conventional method for grading oil palm fresh fruit bunches (FFBs) involves workers assessing the condition of the FFBs visually by making a small cut in the fruit to see the mesocarp colour and counting the number of loosened fruits per FFB. However, this manual grading process is time- consuming, labour-intensive and prone to biased appraisal and human error. These issues can lead to higher harvesting and production costs. Therefore, there is a need for a rapid, reliable and accurategrading technique to detect oil palm FFB ripeness. Successful automation of the process requires asystem that can produce results comparable to human grading. [Source: Oil palm fruit grading using a hyperspectral device and machine learning algorithm (O.M. Bensaeed et. al.) [IOP Conf. Series: Earth and Environmental Science 20 (2014) 0102017]]

[0002] SUMMARY OF INVENTION The present invention relates generally to a method for detecting, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, estimating weight, grading and mapping of a predetermined object. More specifically, the present invention relates to a method and system for detecting, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, grading, estimating weight and grading of oil palm fresh fruit bunches based on a first classification, a second classification and a third classification.Accordingly, the present invention provides a computer-implemented method for detecting, grabbingand picking up, counting, obtaining a geolocation, annotating, classifying, estimating weight, grading and mapping of at least one predetermined object in real-time mode, the method comprising a) obtaining or capturing a plurality of visual data from a surrounding area where an at least one predetermined object is present using at least one imaging device (101) whereby the plurality of visual data is displayed on a screen or display panel (102) in a movable apparatus (103), b) transmitting the plurality of visual data to an at least one processing unit (104) of a computing device, c) identifying and detecting the at least one predetermined object by the at least one processing unit (104) trained with at least one AI machine learning algorithm using the plurality of visual data, d) transmitting at least one signal from the at least one processing unit (104) to an at least one grabbing means (105), e) moving of the at least one grabbing means (105) toward a direction of the at least one predetermined object, f) grabbing and picking of the at least one predetermined object by the at least one grabbing means (105), g) obtaining at least one geolocation of the at least one predetermined object by way of at least one navigation system (106) coupled with the at least one AI machine learning algorithm, h) counting the at least one predetermined object by the at least one AI machine learning algorithm and i) annotating, classifying, estimating weight and grading of the at least one predetermined object by an at least one classification of the at least one AI machine learning algorithm via an at least one training dataset comprising the plurality of visual data. Further, the present invention provides a system for detecting, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, estimating weight, grading and mapping of at least one predetermined object in real-time mode, the system comprising a) obtaining or capturing a plurality of visual data from a surrounding area where an at least one predetermined object is present using at least one imaging device (101) whereby the plurality of visual data is displayed on a screen or display panel (102) in a movable apparatus (103), b) transmitting the plurality of visual data to an at least one processing unit (104) of a computing device, c) identifying and detecting the at least one predetermined object by the at least one processing unit (104) trained with at least one AI machine learning algorithm using the plurality of visual data, d) transmitting at least one signal from the at least one processing unit (104) to an at least one grabbing means (105), e) moving of the at least one grabbing means (105) toward a direction of the at least one predetermined object, f) grabbing and picking of the at least one predetermined object by the at least one grabbing means (105), g) obtaining at least one geolocation of the at least one predetermined object by way of at least one navigation system (106) coupled with the at least one AI machine learning algorithm, counting the at least one predetermined object by the at least one AI machine learning algorithm and annotating, classifying, estimating weight and grading of the at least one predetermined object by an at least one classification of the at least one AI machine learning algorithm via an at least one training dataset comprising the plurality of visual data.

[0003] BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 illustrates the conventional process of harvesting oil palm FFBs from oil palm estates to palm oil mills using manual grading. Figure 2 illustrates generally the harvesting of oil palm FFBs from oil palm estates to palm oil mills via the present invention. Figure 3 shows five images which illustrates the components or parts of the present invention whichare: at least one imaging device (101), a screen or display panel (102), a movable apparatus (103), atleast one processing unit (104), at least one grabbing means (105), at least one navigation system (106), a machine-readable storage medium (107), a network (108) and at least one mapping software (109). Figure 4 illustrates the geolocation preciseness in comparison to a Trimble GPS. To verify the accuracy of the GPS data, a comparison was conducted between 50 location data points obtained from the invention and a Trimble GPS. Figure 5 illustrates a live display on the screen (102) of the present invention. Figure 6 illustrates figures obtained from individual frames from the cameras used for the purposes of the present invention. Figure 7 illustrates the architecture of present invention. Figure 8 illustrates old rodent damaged oil palm FFBs vs, new rodent damaged oil palm FFBs. Figure 9 illustrates oil palm mapping of the oil palm FFBs using data from present invention. Figure 10 illustrates the oil palm FFB density mapping using data of the present invention. Figure 11 illustrates the ripeness mapping of the oil palm FFBs using FFB counting data of the present invention. Figure 12 illustrates FFB rodent damage mapping using data from the present invention. Figure 13 illustrates oil palm FFB yield mapping based on estimated weight data collected for the oil palm FFBs of the present invention. . DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS OF THE PRESENT INVENTION The present invention relates generally to a method for detecting, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, estimating weight, grading and mapping of a predetermined object. More specifically, the present invention relates to a method and system for detecting, grabbing, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, grading, estimating weight and grading of oil palm fresh fruit bunches (FFBs) based on a first classification, a second classification and a third classification.It should be noted that the parameters for the present invention were not obvious and have beendetermined by the inventors after conducting numerous trials, observations, discussions, and combining their expertise. It would have required significant effort and analysis to determine these parameters. It is important to note that none of the prior arts listed and referred to above specifically describe the method and system of this present invention. At an oil palm mill, the conventional method of grading the oil palm FFBs involves an experienced worker stationed at the mill ramp who observes, checks, and grades the FFBs based on their level of ripeness and damage caused by rodents, per Figure 1. Figure 1 illustrates the conventional process of harvesting oil palm FFBs from oil palm estates to palm oil mills using manual grading. Tracing the original location of oil palm FFBs is not possible through conventional means. Moreover, the consistency and accuracy of the grading process varies between workers, as each grader can have different experiences, leniency, attitude and perspectives when grading the FFBs at the mill ramp. The data is manually recorded in logbooks, which require storage space and are prone to errors. The integrity of the grader plays a significant role in the accuracy of the recorded data. Additionally, the conventional means of grading FFBs is not efficient, optimized, or accurate in determining the ripeness levels and rodent damage of the oil palm FFBs. Overall, the present invention offers several benefits, including the elimination of human errors, reduced manpower, and optimized operating costs. It allows for on-site, online, and automated grading and weight estimation of oil palm FFBs in the estates, which makes it possible to trace the quality of oil palm FFBs. Moreover, it provides comprehensive data on the harvested oil palm FFBs, which can be automatically stored digitally. Estate management can focus their efforts on areas with immediate issues, and rodent damage data can be obtained faster (at least 1 census per month) compared to conventional manual census systems, which are only done twice per year. The objectives of the present invention are as follows: A first object of the present invention is to offer a method and system that is effective, efficient, optimized, fast, accurate and targeted in the identification, detection, grabbing, and counting of oil palm FFBs on-site. The present invention also provides real-time geolocation which allows for tracing of the oil palm FFBS to its original location and grading of the FFBs. A second object of the present invention is to provide a method and system for grading of oil palm FFBs based on a first classification, a second classification and a third classification whereby the firstclassification refers to a ripeness level of the oil palm FFBs, the second classification refers to acondition or state of the oil palm FFBs (normal FFBs vs. rodent damaged FFBs) and the third classification refers to a weight estimation of the oil palm FFBs. The third object of the present invention is to provide a method and system that integrates with the harvesting operations and does not require additional workers to grade the oil palm FFBs for rodent damage. In contrast to conventional methods where trained personnel are sent to conduct a census in the oil palm estates, which requires additional manpower and is time-consuming, the present invention eliminates this need. A fourth object of the present invention is to provide an automated, accurate, and efficient method for counting oil palm FFBs. Currently, FFB counting is carried out manually by the operator of the movable apparatus in the estates, which is dependent on the integrity of the worker. This present invention aims to eliminate the need for manual counting and improve accuracy and efficiency. A fifth object of the present invention is to provide a mapping means that uses colour-coding to grade oil palm FFBs based on their characteristics, such as their ripeness level, estimated weight, and the extent of rodent damage. By using this mapping system, estate management will be able to develop more precise and targeted approaches to maintain the health of oil palm trees and improve yields. This information will also help them identify areas with rodent damage and take early steps to prevent it from spreading to other areas. Traditionally, the assessment of rodent damage is done twice a year, but due to the current labour shortage in the industry, it has become a challenge. The top three estates of Sime Darby Plantation with the worst rodent damage have suffered a loss of around RM250,000 to RM300,000 per year. To estimate the FFB weight, the R&D team usually conducts yield recording whenever a new technology or product that can have an impact on FFB yield / weight is tested in an estate. They compare the yield of the area tested with the technology to that of a controlled area. Traditionally, this practice requires four people and manual weight measuring devices. However, due to the increasing labour shortage, this method is no longer feasible. This invention provides a newer and more modern way of performing oil palm FFB weight / yield recording. A sixth object of the present invention is to eliminate the need for manual census to determine rodent damage, allowing manpower to be optimized for other purposes. With the current invention, data on rodent damage, ripeness, and FFB weight can be obtained more frequently and efficiently than through conventional manual means. This enables estate management to focus on other problematic areas related to rodent damage, allowing for increased productivity. A seventh object of the present invention is to provide a means to trace the origin of counted oil palm FFBs.To the best of the inventors' knowledge and based on available prior art, there is currently no knownmethod or system that is similar to the present invention in the oil palm industry. The proposed system is a significant breakthrough invention in the industry. The method and system that focuses on annotating, classifying, and grading based on the condition of oil palm FFBs (whether they are normal or damaged by rodents) and estimating the weight of oil palm FFBs is ground-breaking and novel. This method has not been used in the industry to date and is considered a significant development. Overall, the present invention which focuses on detecting, grabbing, counting, obtaining geolocation, annotating, classifying, estimating weight, grading, and mapping of oil palm fresh fruit bunches (FFBs) based on various classifications, is indeed novel and inventive. It represents a breakthrough in the industry, moving towards mapping of oil palm FFB quality on a per palm basis. The inventors claim that their invention is unique and cannot be predicted by someone knowledgeable in the relevant field. This is because the AI machine learning algorithms require a significant amount of data from the oil palm estates to be collected and analysed. Qualified personnel must verify the collected data and outcomes, and extensive on-site testing is required to establish a method and system that can function effectively in the actual / harsh environment of the oil palm estates. While the present invention is described in detail using illustrative drawings and embodiments, it should be understood that the detailed description is not intended to limit the invention to the embodiments of drawings or drawings described, nor is it intended to limit the invention to the particular form disclosed, but rather to cover all modifications, equivalents, and alternatives falling within the scope of the present invention. The present invention is described in various embodiments with reference to accompanying figures, where reference numerals in the drawing correspond to the features specified in the description. However, the present invention can take many various forms and should not be interpreted as restricted to the examples described herein. As a result, embodiments are described so that this disclosure is thorough and complete, and that those knowledgeable in the art can fully understand the scope of the invention. The numerical values, ranges, and materials described in the comprehensive description are supplied as examples only and are not intended to limit the scope of the present invention's claims. The terminology and phraseology used herein are only for descriptive reasons and are not meant to be restrictive in scope. Words like "including", "comprising", "having", "containing" or "involving" and other variations are meant to be broad and embrace the subject matter as specified, including equivalents and extra subject matter not recited, such as other components or steps. The details of the present invention will now be described in relation to the accompanying Figures 2 to13:The present invention provides a computer-implemented method for detecting, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, estimating weight, and grading of at least one predetermined object in real-time mode, the method comprising: a) obtaining or capturing a plurality of visual data from a surrounding area where an at least one predetermined object is present using at least one imaging device (101) whereby the plurality of visual data is displayed on a screen or display panel (102) in a movable apparatus (103); b) transmitting the plurality of visual data to an at least one processing unit (104) of a computing device; c) identifying and detecting the at least one predetermined object by the at least one processing unit (104) trained with at least one AI machine learning algorithm using the plurality of visual data; d) transmitting at least one signal from the at least one processing unit (104) to an at least one grabbing means (105); e) moving of the at least one grabbing means (105) toward a direction of the at least one predetermined object; f) grabbing and picking of the at least one predetermined object by the at least one grabbing means (105); g) obtaining at least one geolocation of the at least one predetermined object by way of at least one navigation system (106) coupled with the at least one AI machine learning algorithm; h) counting the at least one predetermined object by the at least one AI machine learning algorithm; and i) annotating, classifying, estimating weight and grading of the at least one predetermined object by an at least one classification of the at least one AI machine learning algorithm via an at least one training dataset comprising the plurality of visual data. The method includes use of the movable apparatus (103), any type of movable apparatus (103) depending on the preference and needs of the user of the present invention. The at least one imaging device (101) is located or positioned on a left side and / or right side of the movable apparatus (103). The at least one AI machine learning algorithm is a first AI machine learning algorithm, a second AI machine learning algorithm, a third AI machine learning algorithm and fourth AI machine learning algorithm. The first AI machine learning algorithm generates specific instructions to identify, detect and count the at least one predetermined object. The first AI machine learning algorithm generates specific instructions to identify and detect the at least one grabbing means (105). Two intersecting lines (B) are displayed on the screen or display panel (102), an at least one boundary box (A) or grid is generated around the at least one predetermined object and an at least one boundary box (A) or grid is generated around the at least one grabbing means (105) when captured by the at least one imaging device (101). Step f) takes place when the at least one boundary box (A) or grid of the at least one predetermined object and the at least one boundary box (A) or grid of the at least one grabbing means (105) overlaps and intersects one of the two intersecting lines (B). A global positioning system (GPS) coordinates and timestamps are obtained for step g) and displayed on the screen or display panel (102). The at least one classification is a first classification, a second classification and a third classification. The at least one training dataset comprises a first training dataset, a second training dataset and a third training dataset. The at least one AI machine learning algorithm of step i) is the second AI machine learning algorithm which generates specific instructions to annotate, classify, estimate weight and grade the at least one predetermined object as contained in the at least one boundary box (A) by the first classification. The first classification refers to a ripeness level of the at least one predetermined object as trained by the first training dataset. The at least one AI machine learning algorithm of step i) is the third AI machine learning algorithm which generates specific instructions to annotate, classify, estimate weight and grade the at least one predetermined object as contained in the at least one boundary box (A) by the second classification. The second classification refers to a condition or state of the at least one predetermined object as trained by the second training dataset. The at least one AI machine learning algorithm of step i) is the fourth AI machine learning algorithm which generates specific instructions to annotate, classify, estimate weight and grade the at least one predetermined object as contained in the at least one boundary box (A) by the third classification. The third classification refers to a weight estimation of the at least one predetermined object as trained by the third training dataset. The first classification is further divided into at least one first sub-classification. The at least one first sub-classification refers to the at least one predetermined object which is under- ripe, unripe, ripe, overripe and / or empty. The second classification is further divided into at least one second sub-classification. The at least one second sub-classification refers to a normal predetermined object (not damaged by rodents) or a predetermined object as damaged by rodents. The at least one predetermined object means oil palm fresh fruit bunches (FFBs). The oil palm FFBs are identified based on at least one predetermined feature or characteristic. The at least one predetermined feature or characteristic is appearance, shape, colour, size, texture, number of empty sockets and / or in any combination thereof. The method further includes a machine-readable storage medium (107) in the movable apparatus (103) for automatically storing data on the count, at least one geolocation, annotation, classification, weight estimation and grading of the at least one predetermined object. The data is transmitted by way of a network (108) to an online storage coupled to an at least one processor to provide a set of instructions to an at least one mapping software (109). The at least one mapping software (109) is to generate at least one map based on the GPS coordinates, timestamp, classification, weight estimation and grading of the at least one predetermined object. The at least one map is a first map, a second map and a third map. The first map is to illustrate or map the at least one predetermined object by way of the at least one first sub-classification. The second map is to illustrate or map the at least one predetermined object by way of the at least onesecond sub-classification.The third map is to illustrate or map the at least one predetermined object by way of the third classification. The method further comprises use of an at least one sensor whereby the at least one sensor is configured to detect light intensity in the surrounding area. The at least one light is switched on when the light intensity reduces below a set threshold level or the at least one light is switched off when the light intensity reaches the set threshold level. The at least grabbing means (105) is an at least one grabber arm. The at least one grabber arm is controlled manually, automated or autonomous. The at least one grabber arm moves horizontally and / or vertically and is rotatable 180°, either clockwise or anti-clockwise. The least one imaging device (101) is a camera system. The at least one software is an online information mapping software. The method is able to achieve at least 80% accuracy for the counting, grading via the first, second and third classifications of the at least one predetermined object. The present invention also provides a system for detecting, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, estimating weight and grading of at least one predetermined object in real-time mode, the system comprising: a) obtaining or capturing a plurality of visual data from a surrounding area where an at least one predetermined object is present using at least one imaging device (101) whereby the plurality of visual data is displayed on a screen or display panel (102) in a movable apparatus (103); b) transmitting the plurality of visual data to an at least one processing unit (104) of a computing device;c) identifying and detecting the at least one predetermined object by the at least one processingunit (104) trained with at least one AI machine learning algorithm using the plurality of visual data; d) transmitting at least one signal from the at least one processing unit (104) to an at least one grabbing means (105); e) moving of the at least one grabbing means (105) toward a direction of the at least one predetermined object; f) grabbing and picking of the at least one predetermined object by the at least one grabbing means (105); g) obtaining at least one geolocation of the at least one predetermined object by way of at least one navigation system (106) coupled with the at least one AI machine learning algorithm; h) counting the at least one predetermined object by the at least one AI machine learning algorithm; and i) annotating, classifying, estimating weight and grading of the at least one predetermined object by an at least one classification of the at least one AI machine learning algorithm via an at least one training dataset comprising the plurality of visual data. The system includes use of the movable apparatus (103). The at least one imaging device (101) is located or positioned on a left side and / or right side of the movable apparatus (103). The at least one AI machine learning algorithm is a first AI machine learning algorithm, a second AI machine learning algorithm, a third AI machine learning algorithm and a fourth AI machine learning algorithm. The first AI machine learning algorithm generates specific instructions to identify, detect and count the at least one predetermined object. The first AI machine learning algorithm generates specific instructions to identify and detect the at least one grabbing means (105). Two intersecting lines (B) are displayed on the screen or display panel (102), an at least one boundary box (A) or grid is generated around the at least one predetermined object and an at least one boundary box (A) or grid is generated around the at least one grabbing means (105) when captured by the at least one imaging device (101). Step (f) takes place when the at least one boundary box (A) or grid of the at least one predetermined object and the at least one boundary box (A) or grid of the at least one grabbing means (105) overlaps and intersects one of the two intersecting lines (B). The global positioning system (GPS) coordinates and timestamps are obtained for step g) and displayed on the screen or display panel (102). The at least one classification is a first classification, a second classification and a third classification. The at least one training dataset comprises a first training dataset, a second training dataset and a third training dataset. The at least one AI machine learning algorithm of step i) is the second AI machine learning algorithm which generates specific instructions to annotate, classify, estimate weight and grade the at least one predetermined object as contained in the at least one boundary box (A) by the first classification. The first classification refers to a ripeness level of the at least one predetermined object as trained by the first training dataset. The at least one AI machine learning algorithm of step i) is the third AI machine learning algorithm which generates specific instructions to annotate, classify, estimate weight and grade the at least one predetermined object as contained in the at least one boundary box (A) by the second classification. The second classification refers to a condition or state of the at least one predetermined object as trained by the second training dataset. The at least one AI machine learning algorithm of step i) is the fourth AI machine learning algorithm which generates specific instructions to annotate, classify, estimate weight and grade the at least one predetermined object as contained in the at least one boundary box (A) by the third classification. The third classification refers to a weight estimation of the at least one predetermined object as trained by the third training dataset. The first classification is further divided into at least one first sub-classification. The at least one first sub-classification refers to the at least one predetermined object which is under-ripe, unripe, ripe, overripe and / or empty. The second classification is further divided into at least one second sub- classification. The at least one second sub-classification refers to a normal predetermined object (not damaged by rodents) or a predetermined object as damaged by rodents. The at least one predetermined object means oil palm fresh fruit bunches (FFB). The oil palm FFBs are identified based on at least one predetermined feature or characteristic. The at least one predetermined feature or characteristic is appearance, shape, colour, size, texture, number of empty sockets and / or in any combination thereof. The system further includes a machine-readable storage medium (107) in the movable apparatus (103) for automatically storing data on the count, at least one geolocation, annotation, classification, weight estimation and grading of the at least one predetermined object. The data is transmitted by way of a network (108) to an online storage coupled to an at least one processor to provide a set of instructions to an at least one mapping software (109). The at least one mapping software (109) is to generate at least one map based on the GPS coordinates, timestamp, classification, weight estimation and grading of the at least one predetermined object. The at least one map is a first map, a second map and a third map. The first map is to illustrate or map the at least one predetermined object by way of the at least one first sub-classification. The second map is to illustrate or map the at least one predetermined object by way of the at least one second sub-classification. The third map is to illustrate or map the at least one predetermined object by way of the third classification. The system further comprises use of an at least one sensor whereby the at least one sensor is configured to detect light intensity in the surrounding area. The at least one light is switched on when the light intensity reduces below a set threshold level or the at least one light is switched off when the light intensity reaches the set threshold level. The system is able to achieve at least 80% accuracy for the counting, grading via the first,second and third classifications of the at least one predetermined object.Findings and further details of the present invention: The at least one predetermined object means the oil palm FFBs. The at least one classification is a first classification, a second classification and a third classification. The AI algorithms are a first AI machine learning algorithm, a second AI machine learning algorithm, a third AI machine learning algorithm and a fourth AI machine learning algorithm. The at least one training dataset comprises a first training dataset, a second training dataset and a third training dataset. Once the training of the dataset was completed, the AI machine algorithm underwent testing. The obtained results from the training were compared to the conventional / manual method of grading oil palm FFBs by a skilled worker to verify the accuracy of the algorithm. In cases where it was necessary, additional data was collected or obtained from the oil palm estates to enhance the accuracy of the trained AI algorithms. Based on current testing by the inventors, the present invention achieves the following accuracy: No Item Training dataset Accuracy 1 Counting of oil palm FFBs 36,435 92.6% 2 First classification 286,216 61.1% 3 Second classification 179,068 76.7% 4 Third classification 16,385 70.0% Table 1 The present invention is on target to achieve at least 80% accuracy (and more) for the counting, grading via the first, second and third classifications of the at least one predetermined object. The present invention achieves at least 61.1% accuracy for the first classification whereby the percentage on accuracy is obtained through validation using manual grading. The first classification refers to a ripeness level of the oil palm FFBs as trained by the first training dataset. The present invention achieves at least 76.7% accuracy for the second classification whereby the percentage on accuracy is obtained through validation using manual grading. The second classification refers to a condition or state of the oil palm FFBs as trained by the second training dataset. The present invention achieves at least 70.0% accuracy for the third classification whereby thepercentage on accuracy is obtained through validation using manual grading.The third classificationrefers to a weight estimation as trained by the third training dataset.The present invention achieves at least 92.6% accuracy for the counting of the oil palm FFBs whereby the percentage on accuracy is obtained through validation using manual grading. The method and system used in oil palm estates have been proven and tested to achieve high accuracy in the counting and grading of predetermined objects. At least 92.6% accuracy has been achieved in counting the objects, while the first, second and third classifications have achieved grading accuracies of at least 61.1%, 76.7% and 70.0% respectively. These percentages are obtained through validation using manual grading. The accuracy percentage is increasing every day as more dataset is being trained using an AI machine learning algorithm. The system's detection accuracy is expected to improve continuously as it already works in the estates for the present invention. It is expected that the accuracy of the percentage will reach at least 80% for the first, second, and third classifications in due time. The inventors would like to file a patent application to secure the novelty of the present invention while continuing to train more datasets. The present invention relates to a method and system for detecting, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, estimating weight, grading and mapping of oil palm fresh fruit bunches based on a first classification, a second classification and a third classification, which comprises the following steps: i. data collection; ii. data classification; iii. data annotation; iv. training of algorithm; and v. testing and validating of algorithm. The first AI machine learning algorithm generates specific instructions to identify, detect and count the The first AI machine learning algorithm also generates specific instructions to identify and detect the at least one grabbing means (105). The oil palm FFBs are identified based on at least one predetermined feature or characteristic. The at least one predetermined feature or characteristic is appearance, shape, colour, size, texture, number of empty sockets and / or in any combination thereof. With respect to the first AI machine learning algorithm for counting the oil palm FFBs, two lines intersecting (B) each other on a plane are displayed on the screen or display panel (102) in the movable apparatus (103). An imaging device (101) captures the grabber arm and each oil palm FFB, generatinga boundary box (A) or grid for each. Counting only starts when the boundary box (A) or grid containingan oil palm FFB overlaps with the grabber arm's boundary box (A) or grid and crosses one of the two intersecting lines (B) on the display panel (102). This method provides an accurate count of oil palm FFBs, as only those picked up by the grabber arm are counted, to avoid over-counting the uncollected FFBs. The first classification is further divided into at least one first sub-classification. The at least one first sub-classification refers to the under-ripe FFBs, unripe FFBs, ripe FFBs, overripe FFBs and / or empty bunches. The second AI machine learning algorithm which generates specific instructions to annotate, classify, and grade the oil palm FFBs as contained in the at least one boundary box (A) by the first classification (oil palm FFB ripeness level). The second classification is further divided into at least one second sub-classification. The at least one second sub-classification refers to a normal predetermined object (not damaged by rodents) or a predetermined object as damaged by rodents (i.e. normal oil palm FFBs or oil palm FFBs damaged by rodents).The third AI machine learning algorithm generates specific instructions to annotate, classify, and grade the oil palm FFBs as contained in the at least one boundary box (A) by the second classification. The fourth AI machine learning algorithm generates specific instructions to annotate, classify, estimate weight and grade the oil palm FFBs as contained in the at least one boundary box (A) by the third classification. The least one imaging device (101) is a camera system, which can be any type of camera depending on the needs / preference of the user of the present invention. Whenever the movable apparatus (103) captures images using at least one camera positioned on the left or right side, boundary boxes (A) will be generated around the oil palm FFBs and the grabber arm. The oil palm FFBs need to be identified and detected in order for the grabber arm to pick them up. Once this is complete, the first AI machine learning algorithm will count the oil palm FFBs. The movable apparatus (103) displays multiple visual data such as images and video feeds of oil palm FFBs on the screen or display panel (102). The operator can view these in real-time mode, per Figure 5 which illustrates a live display on the screen (102) of the present invention. A navigation system (106), specifically a global positioning system (GPS), is connected to the first AI machine learning algorithm to obtain the geolocation of the oil palm FFBs when it is picked up by the grabber arm. The GPS system provides the coordinates and timestamps which are displayed on the movable apparatus's (103) display panel (102). The geolocation is obtained before counting of the oil palm FFBs begin. The GPS geolocation receiver captures the precise location and time when a grabber arm picks up an oil palm FFB. To verify the accuracy of the GPS data, a comparison was conducted between 50 location data points obtained from the invention and a Trimble GPS per Figure 4. The analysis revealed a difference of about 3.17 meters, which is considered accurate as it is equivalent to the distance between two adjacent palm trees. This demonstrates the effectiveness of the method in providing precise location information of the oil palm FFBs in the estates. The imaging device (101) is a camera system that is installed or positioned on either side of the movable apparatus (103). The number of camera systems required depends on the user's preference and needs. For the purposes of this invention, the inventors used the following camera systems: • at least one RealSense 3D camera, which detects, counts, and estimates the weight of the oil palm FFBs; and • at least one 4K 2D dome camera, which grades the oil palm FFBs based on the first and second classifications. Real-time images and / or video feeds of oil palm FFBs can be obtained through online streaming from a RealSense depth camera. Offline data from a RealSense 3D camera or a 4K 2D dome camera can also be used. The video feeds can capture multiple angles of the FFBs, increasing the efficiency of the AI machine-learning algorithms used for grading of the present invention. Figure 6 illustrates figures obtained from individual frames from the cameras used for the purposes of the present invention. The AI machine learning algorithms used in the present invention are trained through a deep learning model that is based on supervised learning techniques. For the purposes of the present invention, the first AI machine learning algorithm used by the inventors is the YOLOv4 (you only look once version 4), a real-time object detection software that identifies, detects and counts oil palm FFBs. The inventors of the present invention used ResNet-50, for the second, third and fourth AI machine learning algorithms which is a 50-layer deep convolutional neural network. These algorithms classify, annotate and grade the oil palm FFBs into several classifications and sub-classifications as listed in Table 2. Classification Classification First classification a) Under ripe (Ripeness) b) Unripe c) Ripe d) Overripe e) Empty bunch Second classification a) normal (undamaged) (Rodent damage) b) rodent damage Third classification FFB weight (Weight estimation) Table 2 The oil palm FFBs are graded based on their classification according to ripeness levels and conditions, which is either rodent damage or normal. Skilled workers manually grade and weigh the FFBs, and thenthey are annotated for the purpose of training and testing the second, third, and fourth AI machinelearning algorithms for grading and weighing. Data annotation is crucial for providing labelled datasets that produce accurate results. The second and third AI machine learning algorithms are trained using a large amount of training datasets, allowing them to accurately grade the oil palm FFBs based on the first, second, and third classifications. Figure 7 illustrates the architecture of present invention. The algorithm is trained and tested multiple times until satisfactory accuracy is reached. Once the data training was completed, the AI machine algorithm underwent testing. The obtained results from the training were compared to the conventional / manual method of grading oil palm FFBs by a skilled worker to verify the accuracy of the algorithm. In cases where it was necessary, additional data was collected or obtained from the oil palm estates to enhance the accuracy of the trained algorithm. The inventors referred to the classification of oil palm FFBs based on their ripeness level, which follows the grading standards used by the Malaysian Palm Oil Board (MPOB) as outlined in Table 3. The grading system takes into account the number of empty sockets in the FFB, as well as its colour, size, and texture. Oil Palm FFB ripeness levels Description Unripe 0 detached fruitlets Under Ripe 1 to 9 sockets from fully developed fruitlets Ripe ≥10 sockets from fully developed fruitlets Over Ripe >50% sockets of fruitlets detached from the whole bunch or at least 10% fruitlets still attached to the bunch Empty Bunch >90% fruitlets detached from the whole bunch Table 3 Figure 8 illustrates old rodent damaged oil palm FFBs vs, new rodent damaged oil palm FFBs. Old rodent damaged oil palm FFBs refers to FFBs in which the fruitlets have been eaten completely, leaving only the shell visible on the inside. On the other hand, new rodent damaged oil palm FFBs refers to FFBs in which the fruitlets have been partially eaten, making it impossible to see the shell on the inside. The AI machine learning algorithm can easily identify these damages through the visible holes. The inventors of this invention focus on the classification of oil palm FFBs that are normal versus those that have been damaged by rodents. However, they only consider the damage caused by old rodent infestations, and not new ones. The reason for this is that new damage caused by rodents looks very similar to damage caused by cutting or falling off from the oil palm trees. Old rodent damage refers to FFBs in which the fruitlets have been eaten completely, leaving only the shell visible on the inside. On the other hand, new rodent damage refers to FFBs in which the fruitlets have been partially eaten, making it impossible to see the shell on the inside. The system includes a machine-readable storage medium, such as a solid-state drive (SSD), for storing data on the count, geolocation, annotation, classification, and grading of the oil palm FFBs. The data is transmitted through a network to an online storage, where it can be accessed by at least one mappingsoftware, such as a geographic information mapping system. The mapping software generates at leastone map based on the GPS coordinates, timestamp, classification, weight estimation, and grading of the oil palm FFBs. An online information mapping system produces three maps to illustrate various characteristics of a predetermined object. • The first map depicts the object's at least one first sub-classification based on one or more predetermined features or characteristics. This map is used to create the FFB ripeness mapping.• The second map illustrates the object's at least one second sub-classification based on one or morepredetermined features or characteristics. This map is used to create the FFB rodent damages mapping. • The third map displays the object's at least one third classification based on one or more predetermined features or characteristics. This map is used to create the FFB yield mapping. The method and system presented in this invention can be installed on any mobile apparatus, making it suitable for use in coastal and inland properties, depending on the user's needs and preferences. Any type of grabber arm can be utilized for the purposes of this invention. The system is highly modular, allowing it to be easily attached to any trolley or moving object, enabling easy maneuverability in estates, mills, and other similar environments. Oil palm mapping can be easily accomplished with any mapping software, such as ArcGIS, Google Maps, or others that are feasible to use. Figure 9 illustrates oil palm mapping of the oil palm FFBs using data from present invention. Each dot represents an oil palm FFB collected from an oil palm estate with a mechanical tractor grabber as the movable apparatus (103) of the present invention. The same dataset can be used to generate a FFB density analysis, as shown in Figure 10. Figure 10 illustrates the oil palm FFB density mapping using data of the present invention Figure 11 illustrates the ripeness mapping of the oil palm FFBs using FFB counting data of the present invention. Each dot represents an oil palm FFB collected from an oil palm estate and is classified into different ripeness levels, represented by various colours. The ripeness levels are as follows: a) Under-ripe b) Unripe c) Ripe d) Overripe e) Empty bunch. Figure 12 which illustrates FFB rodent damage mapping using data from the present invention.The image in Figure 12 shows a mapping of oil palm fresh fruit bunches (FFBs) based on two factors: their weight and the extent of damage caused by rodents. Each dot in the image represents an FFB that was collected from an oil palm estate and is classified based on its rodent damage and weight. The dots are coloured differently to show the following: • Normal oil palm FFBs; and • Rodent-damaged oil palm FFBs. Figure 13 illustrates oil palm FFB yield mapping based on estimated weight data collected for the oil palm FFBs of the present invention. Each dot represents oil palm FFBs collected from an oil palm plantation. The FFBs are classified by weight, represented by different colours as follows: • FFBs weighing between 30.60kg and 47.91 kg; • FFBs weighing between 16.14kg and 30.59 kg; and • FFBs weighing between 4.05kg and 16.13 kg. This mapping approach allows for effective estate management, especially when it comes to managing rat problems. It enables estate managers to identify which areas require rat baits and which ones do not, which leads to cost savings. Additionally, the mapping allows for frequent censuses on rat damage. This means that target areas can be quickly identified and fixed, which is better than the conventional method of conducting censuses only twice a year. The method and system are suitable for use in any soil and estate conditions. The method and system used in oil palm estates have been proven and tested to achieve high accuracy in the counting and grading of predetermined objects. At least 92.6% accuracy has been achieved in counting the objects, while the first, second and third classifications have achieved grading accuracies of at least 61.1%, 76.7% and 70.0% respectively. These percentages are obtained through validation using manual grading. The accuracy percentage is increasing every day as more dataset is being trained using an AI machine learning algorithm. The system's detection accuracy is expected to improve continuously as it already works in the estates for the present invention. It is expected that the accuracy of the percentage will reach at least 80% for the first, second, and third classifications in due time. The inventors would like to file a patent application to secure the novelty of the present invention while continuing to train more datasets. Summary While there are other prior arts related to ripeness classification, the method and system that focuseson annotating, classifying, and grading based on the condition of oil palm FFBs (whether they arenormal or damaged by rodents) and estimating the weight of oil palm FFBs is ground-breaking and novel. This method has not been used in the industry to date and is considered a significant development. The invention, which focuses on detecting, grabbing, counting, obtaining geolocation, annotating, classifying, estimating weight, grading, and mapping of oil palm fresh fruit bunches (FFBs) based on various classifications, is indeed novel and inventive. It represents a breakthrough in the industry, moving towards mapping of oil palm FFB quality on a per palm basis. To the best of the inventors' knowledge and based on available prior art, there is currently no known method or system that is similar to the present invention in the oil palm industry. The proposed system is a significant breakthrough invention in the industry. The inventors claim that their invention is unique and cannot be predicted by someone knowledgeable in the relevant field. This is because the AI machine learning algorithms require a significant amount of data from the oil palm estates to be collected and analysed. Qualified personnel must verify the collected data and outcomes, and extensive on-site testing is required to establish a method and system that can function effectively in the actual / harsh environmentof the oil palm estates.None of the prior art references mentioned and cited above specifically describe the method and system of the present invention. Furthermore, it is not evident from the information and documents cited above that experts in the related field could easily determine the method and system of the present invention. The inventors of the method and system have conducted extensive tests, trials, observations, and discussions with their combined expertise and experience in this field. The parameters and / or combination of the method and system could not be easily determined through mere review of prior art documents in this field of interest. Overall, the present invention offers several benefits, including the elimination of human errors, reduced manpower, and optimized operating costs. It allows for on-site, online, and automated grading and weight estimation of oil palm FFBs in the estates, which makes it possible to trace the quality of oil palm FFBs. Moreover, it provides comprehensive data on the harvested oil palm FFBs, which can be automatically stored digitally. Estate management can focus their efforts on areas with immediate issues, and rodent damage data can be obtained faster (at least 1 census per month) compared to conventional manual census systems, which are only done twice per year. The inventors have determined the parameters of the present invention through a series of trials, observations, and discussions with experts in the field. These parameters and / or their combinations could not have been determined without significant effort, testing, and analysis. Therefore, to the best of the inventors' knowledge, the present invention is novel and innovative. Various modifications to these embodiments as described herein are apparent to those skilled in the art from the description and the accompanying drawings. The description is not intended to be limited to these embodiments as shown with the accompanying drawings but is to provide the broadest scope possible as consistent with the novel and inventive features disclosed. Accordingly, the invention is anticipated to hold on to all other such alternatives, modifications and variations that fall within the scope of the present invention and appended claims.

Claims

CLAIMS 1. A computer-implemented method for detecting, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, estimating weight and grading of at least one predetermined object in real-time mode, the method comprising: a) obtaining or capturing a plurality of visual data from a surrounding area where an at least one predetermined object is present using at least one imaging device (101) whereby the plurality of visual data is displayed on a screen or display panel (102) in a movable apparatus (103); b) transmitting the plurality of visual data to an at least one processing unit (104) of a computing device; c) identifying and detecting the at least one predetermined object by the at least one processing unit (104) trained with at least one AI machine learning algorithm using the plurality of visual data; d) transmitting at least one signal from the at least one processing unit (104) to an at least one grabbing means (105); e) moving of the at least one grabbing means (105) toward a direction of the at least one predetermined object; f) grabbing and picking of the at least one predetermined object by the at least one grabbing means (105); g) obtaining at least one geolocation of the at least one predetermined object by way of at least one navigation system (106) coupled with the at least one AI machine learning algorithm; h) counting the at least one predetermined object by the at least one AI machine learning algorithm; and i) annotating, classifying, estimating weight and grading of the at least one predetermined object by an at least one classification of the at least one AI machine learning algorithm via an at least one training dataset comprising the plurality of visual data.

2. The method of Claim 1, wherein the method includes use of the movable apparatus (103).

3. The method of Claim 1, wherein the at least one imaging device (101) is located or positioned on a left side and / or right side of the movable apparatus (103).

4. The method of Claim 1, wherein the at least one AI machine learning algorithm is a first AI machine learning algorithm, a second AI machine learning algorithm, a third AI machine learning algorithm and a fourth AI machine learning algorithm.

5. The method of Claim 4, wherein the first AI machine learning algorithm generates specific instructions to identify, detect and count the at least one predetermined object.

6. The method of Claim 4, wherein the first AI machine learning algorithm generates specific instructions to identify and detect the at least one grabbing means (105).

7. The method of Claim 1, wherein two intersecting lines (B) are displayed on the screen or display panel (102), an at least one boundary box (A) or grid is generated around the at least one predetermined object and an at least one boundary box (A) or grid is generated around the at least one grabbing means (105) when captured by the at least one imaging device (101).

8. The method of Claim 1, wherein step f) takes place when the at least one boundary box (A) or grid of the at least one predetermined object and the at least one boundary box (A) or grid of the at least one grabbing means (105) overlaps and intersects one of the two intersecting lines (B).

9. The method of Claim 1, wherein global positioning system (GPS) coordinates and timestamps are obtained for step g) and displayed on the screen or display panel (102).

10. The method of Claim 1, wherein the at least one classification is a first classification, a second classification and a third classification.

11. The method of Claim 1, wherein the at least one training dataset comprises a first training dataset, a second training dataset and a third training dataset.

12. The method of Claim 4, wherein the at least one AI machine learning algorithm of step i) is the second AI machine learning algorithm which generates specific instructions to annotate, classify, estimate weight and grade the at least one predetermined object as contained in the at least one boundary box (A) by the first classification.

13. The method of Claim 12, wherein the first classification refers to a ripeness level of the at least one predetermined object as trained by the first training dataset.

14. The method of Claim 4, wherein the at least one AI machine learning algorithm of step i) is the third AI machine learning algorithm which generates specific instructions to annotate, classify, estimate weight and grade the at least one predetermined object as contained in the at least one boundary box (A) by the second classification.

15. The method of Claim 14, wherein the second classification refers to a condition or state of the at least one predetermined object as trained by the second training dataset.

16. The method of Claim 4, wherein the at least one AI machine learning algorithm of step i) is the fourth AI machine learning algorithm which generates specific instructions to annotate, classify, estimate weight and grade the at least one predetermined object as contained in the at least one boundary box (A) by the third classification.

17. The method of Claim 16, wherein the third classification refers to a weight estimation of the atleast one predetermined object as trained by the third training dataset.

18. The method of Claim 12, wherein the first classification is further divided into at least one first sub-classification.

19. The method of Claim 18, wherein at least one first sub-classification refers to the at least one predetermined object which is under-ripe, unripe, ripe, overripe and / or empty.

20. The method of Claim 14, wherein the second classification is further divided into at least one second sub-classification.

21. The method of Claim 20, wherein the at least one second sub-classification refers to a normal predetermined object (not damaged by rodents) or a predetermined object damaged by rodents.

22. The method of Claim 1, wherein the at least one predetermined object means oil palm fresh fruit bunches (FFBs).

23. The method of Claim 22, wherein the oil palm FFBs are identified based on at least one predetermined feature or characteristic.

24. The method of Claim 23, wherein the at least one predetermined feature or characteristic is appearance, shape, colour, size, texture, number of empty sockets and / or in any combination thereof.

25. The method of Claim 1, wherein the method further includes a machine-readable storage medium (107) in the movable apparatus (103) for automatically storing data on the count, at least one geolocation, annotation, classification, weight estimation and grading of the at least one predetermined object.

26. The method of Claim 25, wherein the data is transmitted by way of a network (108) to an online storage coupled to an at least one processor to provide a set of instructions to an at least one mapping software (109).

27. The method of Claim 26, wherein the at least one mapping software (109) is to generate at least one map based on the GPS coordinates, timestamp, classification, weight estimation and grading of the at least one predetermined object.

28. The method of Claim 26, wherein the at least one map is a first map, a second map and a third map.

29. The method of Claim 28, wherein the first map is to illustrate or map the at least onepredetermined object by way of the at least one first sub-classification.

30. The method of Claim 28, wherein the second map is to illustrate or map the at least one predetermined object by way of the at least one second sub-classification.

31. The method of Claim 28, wherein the third map is to illustrate or map the at least one predetermined object by way of the third classification.

32. The method of Claim 1, wherein the method further comprises use of an at least one sensor whereby the at least one sensor is configured to detect light intensity in the surrounding area.

33. The method of Claim 32, wherein an at least one light is switched on when the light intensity reduces below a set threshold level or the at least one light is switched off when the light intensity reaches the set threshold level.

34. A system for detecting, grabbing and picking up, counting, obtaining a geolocation, annotating, classifying, estimating weight and grading of at least one predetermined object in real-time mode, the system comprising: a) obtaining or capturing a plurality of visual data from a surrounding area where an at least one predetermined object is present using at least one imaging device (101) whereby the plurality of visual data is displayed on a screen or display panel (102) in a movable apparatus (103); b) transmitting the plurality of visual data to an at least one processing unit (104) of a computing device; c) identifying and detecting the at least one predetermined object by the at least one processing unit (104) trained with at least one AI machine learning algorithm using the plurality of visual data; d) transmitting at least one signal from the at least one processing unit (104) to an at least one grabbing means (105);e) moving of the at least one grabbing means (105) toward a direction of the at least one predetermined object; f) grabbing and picking of the at least one predetermined object by the at least one grabbing means (105); g) obtaining at least one geolocation of the at least one predetermined object by way of at least one navigation system (106) coupled with the at least one AI machine learning algorithm; h) counting the at least one predetermined object by the at least one AI machine learning algorithm; and i) annotating, classifying, estimating weight and grading of the at least one predetermined object by an at least one classification of the at least one AI machine learning algorithm via an at least one training dataset comprising the plurality of visual data.

35. The system of Claim 34, wherein the system includes use of the movable apparatus (103).

36. The system of Claim 34, wherein the at least one imaging device (101) is located or positioned on a left side and / or right side of the movable apparatus (103).

37. The system of Claim 34, wherein the at least one AI machine learning algorithm is a first AI machine learning algorithm, a second AI machine learning algorithm, a third AI machine learning algorithm and a fourth AI machine learning algorithm.

38. The system of Claim 37, wherein the first AI machine learning algorithm generates specific instructions to identify, detect and count the at least one predetermined object.

39. The system of Claim 37, wherein the first AI machine learning algorithm generates specific instructions to identify and detect the at least one grabbing means (105).

40. The system of Claim 34, wherein two intersecting lines (B) are displayed on the screen or display panel (102), an at least one boundary box (A) or grid is generated around the at least one predetermined object and an at least one boundary box (A) or grid is generated around the at least one grabbing means (105) when captured by the at least one imaging device (101).

41. The system of Claim 34, wherein step f) takes place when the at least one boundary box (A) or grid of the at least one predetermined object and the at least one boundary box (A) or grid of the at least one grabbing means (105) overlaps and intersects one of the two intersecting lines (B).

42. The system of Claim 34, wherein global positioning system (GPS) coordinates and timestamps are obtained for step g) and displayed on the screen or display panel (102).

43. The system of Claim 34, wherein the at least one classification is a first classification, a second classification and a third classification.

44. The system of Claim 34, wherein the at least one training dataset comprises a first training dataset, a second training dataset and a third training dataset.

45. The system of Claim 37, wherein the at least one AI machine learning algorithm of step i) is the second AI machine learning algorithm which generates specific instructions to annotate, classify, estimate weight and grade the at least one predetermined object as contained in the at least one boundary box (A) by the first classification.

46. The system of Claim 45, wherein the first classification refers to a ripeness level of the at least one predetermined object as trained by the first training dataset.

47. The system of Claim 37, wherein the at least one AI machine learning algorithm of step i) is the third AI machine learning algorithm which generates specific instructions to annotate, classify, estimate weight and grade the at least one predetermined object as contained in the at least one boundary box (A) by the second classification.

48. The system of Claim 47, wherein the second classification refers to a condition or state of the at least one predetermined object as trained by the second training dataset.

49. The system of Claim 37, wherein the at least one AI machine learning algorithm of step i) is the fourth AI machine learning algorithm which generates specific instructions to annotate, classify, estimate weight and grade the at least one predetermined object as contained in the at least one boundary box (A) by the third classification.

50. The system of Claim 49, wherein the third classification refers to a weight estimation of the at least one predetermined object as trained by the third training dataset.

51. The system of Claim 46, wherein the first classification is further divided into at least one first sub-classification.

52. The system of Claim 51, wherein at least one first sub-classification refers to the at least one predetermined object which is under-ripe, unripe, ripe, overripe and / or empty.

53. The system of Claim 48, wherein the second classification is further divided into at least one second sub-classification.

54. The system of Claim 53, wherein the at least one second sub-classification refers to a normal predetermined object (not damaged by rodents) or a predetermined object damaged by rodents.

55. The system of Claim 34, wherein the at least one predetermined object means oil palm fresh fruit bunches (FFB).

56. The system of Claim 55, wherein the oil palm FFBs are identified based on at least onepredetermined feature or characteristic.

57. The system of Claim 56, wherein the at least one predetermined feature or characteristic is appearance, shape, colour, size, texture, number of empty sockets and / or in any combination thereof.

58. The system of Claim 34, wherein the system further includes a machine-readable storage medium (107) in the movable apparatus (103) for automatically storing data on the count, at least one geolocation, annotation, classification, weight estimation and grading of the at least one predetermined object.

59. The system of Claim 58, wherein the data is transmitted by way of a network (108) to an online storage coupled to an at least one processor to provide a set of instructions to an at least one mapping software (109).

60. The system of Claim 59, wherein the at least one mapping software (109) is to generate at least one map based on the GPS coordinates, timestamp, classification, weight estimation and grading of the at least one predetermined object.

61. The system of Claim 60, wherein the at least one map is a first map, a second map and a third map.

62. The system of Claim 61, wherein the first map is to illustrate or map the at least one predetermined object by way of the at least one first sub-classification.

63. The system of Claim 61, wherein the second map is to illustrate or map the at least one predetermined object by way of the at least one second sub-classification.

64. The system of Claim 61, wherein the third map is to illustrate or map the at least one predetermined object by way of the third classification.

65. The system of Claim 34, wherein the system further comprises use of an at least one sensor whereby the at least one sensor is configured to detect light intensity in the surrounding area.

66. The system of Claim 65, wherein an at least one light is switched on when the light intensity reduces below a set threshold level or the at least one light is switched off when the light intensity reaches the set threshold level.

Images