System and method for detecting and identifying explosives

Abstract

The present invention relates to a system and method for aerial detection and identification of explosives on the ground, and more particularly to a system and method for detecting and identifying land mines and unexploded ordnance (UXO) on the ground. Additionally, a model is trained to generate orthomosaic images for comparing captured images with training images to predict the location of unknown explosives within a target area.

Description

[Technical Field]

[0001] Related Applications This application claims priority to U.S. Provisional Patent Application No. 63 / 406,234, filed September 13, 2022, the entire contents of which are incorporated herein by reference. [Background technology]

[0002] Landmines and unexploded ordnance (UXO) pose a significant threat to hundreds of thousands of civilians worldwide, significantly reducing their quality of life. Land liberation is the process of detecting all mines and UXO within a specific area, safely removing these objects, and returning explosive-free land to local residents. The most dangerous and time-consuming part of the land liberation and demining process is detecting these objects.

[0003] Despite the efforts of demining organizations and the International Mine Ban Treaty, the number of LUXOs is expected to continue to increase due to ongoing conflicts. In some instances, electromagnetic induction (EMI) and metal detection beep-and-prod methods are used by demining NGOs and national demining operations for mine detection. EMI can be effective in detecting large metal fragments, but often requires very difficult and precise work under very harsh environmental conditions. EMI and metal detection beep-and-prod methods can have several drawbacks, including a high false alarm rate, an inability to detect small, low-metal mines, and / or an inability to detect and clear mines activated by earthquakes.

[0004] An example of a mine that exploits EMI flaws is the Russian-made POM-3. This mine uses a seismic sensor that detonates when a person comes within approximately 16 meters of it, preventing standard detection and destruction procedures. The POM-3 was used during Russia's invasion of Ukraine in 2022. Another mine that exploits flaws in EMI mine detection methods is the PFM-1 antipersonnel mine (see Figure 4). This mine has a plastic casing and a small metal fuse. They can be dropped in large numbers from helicopters and airplanes or fired from mortars. Many PFM-1 mines were used during the former Soviet Union's invasion of Afghanistan. There is also evidence of these mines being used during Russia's invasion of Ukraine in 2022.

[0005] Therefore, there remains a need for improved methods and systems for detecting and identifying explosives such as land mines and unexploded ordnance (UXO) on the ground. Summary of the Invention

[0006] Accordingly, the present invention relates to systems and methods for detecting and identifying explosives on the ground, and more particularly to systems and methods for detecting and identifying land mines and unexploded ordnance (UXO) on the ground.

[0007] Accordingly, one aspect of the present invention is a system for detecting and identifying explosives in a target area using an unmanned aerial vehicle (UAV), comprising: a UAV adapted to include an imaging device suitable for capturing image data and a transmission component capable of transmitting the image data to a control system; 1. An explosives detection and identification processing module including a control system suitable for receiving and processing image data transmitted from a UAV, the module comprising: storing the image data received from the UAV in a machine-readable medium; processing the image data to generate an image of a training area having known explosives; fusing images of the training area, for example using photogrammetry, to generate an orthomosaic image representing the training area with explosives; analyzing one or more segmented images derived from the orthomosaic image using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks (e.g., convolutional neural networks (CNNs)) trained to detect patterns or predefined objects to generate a training model for automating the detection of explosives; identifying and labeling explosives in a training area in the orthomosaic imagery to improve the training model; applying the improved training model to the target area to predict the location of unknown explosives within the target area by comparing the trained orthomosaic image data from the target area with previously generated orthomosaic image data from the training area; an explosives detection and identification processing module including a machine-readable medium having stored thereon instructions for execution by a processor to perform a method including: Including, A system is provided, the system being suitable for detecting and identifying explosives within a target area using an unmanned aerial vehicle (UAV).

[0008] Another aspect of the present invention is an explosives detection and identification processing module including a control system suitable for receiving and processing image data transmitted from an unmanned aerial vehicle (UAV), the module comprising: storing the image data received from the UAV in a machine-readable medium; processing the image data to generate an image of a training area having known explosives; fusing images of the training area, for example using photogrammetry, to generate an orthomosaic image representing the training area with explosives; analyzing one or more segmented images derived from the orthomosaic image using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks (e.g., convolutional neural networks (CNNs)) trained to detect patterns or predefined objects to generate a training model for automating the detection of explosives; identifying and labeling explosives in a training area in the orthomosaic imagery to improve the training model; applying the improved training model to the target area to predict the location of unknown explosives within the target area by comparing the trained orthomosaic image data from the target area with previously generated orthomosaic image data from the training area; a machine-readable medium having stored thereon instructions for execution by a processor to perform a method comprising: An explosives detection and identification processing module is provided, the module being suitable for detecting and identifying explosives in a target area using an unmanned aerial vehicle (UAV).

[0009] Another aspect of the present invention is a method for detecting and identifying explosives in a target area using an unmanned aerial vehicle (UAV), the method comprising: storing the image data received from the UAV in a machine-readable medium; processing the image data to generate an image of a training area having known explosives; fusing images of the training area, for example using photogrammetry, to generate an orthomosaic image representing the training area with explosives; analyzing one or more segmented images derived from the orthomosaic image using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks (e.g., convolutional neural networks (CNNs)) trained to detect patterns or predefined objects to generate a training model for automating the detection of explosives; identifying and labeling explosives in a training area in the orthomosaic imagery to improve the training model; applying the improved training model to the target area to predict the location of unknown explosives within the target area by comparing the trained orthomosaic image data from the target area with previously generated orthomosaic image data from the training area; Including, A method for detecting and identifying explosives is provided, wherein the method is suitable for detecting and identifying explosives in a target area using an unmanned aerial vehicle (UAV). [Brief explanation of the drawings]

[0010] Advantages of the present systems, modules, and associated methods will become apparent from the following detailed description, which should be considered in conjunction with the accompanying drawings, which are not intended to limit the scope of the invention in any way.

[0011] The present disclosure is described with reference to the following figures, in which like numbers are used throughout to refer to like features and components. [Figure 1] FIG. 1 is a schematic diagram of an exemplary system according to the present disclosure. [Figure 2] FIG. 1 is a schematic diagram of an exemplary control system according to the present disclosure. [Figure 3] 1 is an exemplary method of the present disclosure. [Figure 4] 1A-1D are top and side views of an exemplary PFM-1 anti-personnel mine. [Figure 5-7]4 illustrates exemplary image data and a graphical depiction of the method steps of FIG. 3 according to the present disclosure.

[0012] The functional block diagrams, operational sequences, and flow diagrams provided in the figures represent example architectures, environments, and methodologies for implementing novel aspects of the present disclosure. For ease of explanation, the methodologies contained herein may be in the form of functional diagrams, operational sequences, or flow diagrams and may be described as a series of acts; however, it is understood and appreciated that the methodologies are not limited by the order of acts, as some acts may occur in a different order and / or concurrently with other acts than those shown and described herein accordingly. For example, those skilled in the art will understand and appreciate that a methodology could alternatively be represented as a series of interrelated states or events, such as in a state diagram. Furthermore, not all acts depicted in a methodology may be required for novel implementations. DETAILED DESCRIPTION OF THE INVENTION

[0013] The present invention relates to a system and method for detecting and identifying explosives in the ground, and more particularly to a system and method for detecting and identifying land mines and unexploded ordnance (UXO) in the ground.

[0014] The present invention, including systems, processing system modules, and associated methods, will be described for convenience with reference to the following definitions set forth below. Unless otherwise specified, the following terms used herein are defined as follows:

[0015] I. Definition As used herein in the context of the present invention (particularly in the context of the claims), the terms "a," "an," "the," and similar terms are to be construed to cover both the singular and the plural, unless otherwise indicated herein or clearly contradicted by context.

[0016] As used herein, the terms "application programming interface" or "API" are art-recognized and used interchangeably to describe a type of software interface that provides services to other software, i.e., a way for two or more computer programs to communicate with each other. In contrast to a user interface, which connects computers to humans, an application programming interface connects computers to other computers or software to other software. It is not intended for direct use by anyone other than the computer programmers who build it into their software (end users). An API often consists of several different parts that function as tools or services available to programmers. A program or programmer who uses one of these parts is said to call that part of the API. The calls that make up an API are also known as subroutines, methods, requests, or endpoints. An API specification defines these calls; that is, it describes how to use or implement them.

[0017] In this specification, the term "explosives" is used to describe landmines and unexploded ordnance (also known as UXO or LUXO). The term "interfacing" is art-recognized and is used herein to describe a means of communication between two entities, e.g., a system / tool ​​and a user data input. In certain embodiments, interfacing may be bidirectional. In other embodiments, interfacing may be unidirectional. In certain embodiments, such interfacing may be accomplished using a graphical user interface.

[0018] The term "machine-readable medium" is art-recognized and refers to a medium capable of storing data in a form readable by a machine device (rather than by a human). Examples of machine-readable media include magnetic media such as magnetic disks, cards, tapes, and drums; punch cards and paper tapes; optical disks; barcodes; magnetic ink characters; and solid-state devices such as flash-based SSDs. The machine-readable medium of the present invention is non-transitory and therefore does not include signals themselves; i.e., it covers only hardware storage media. Common machine-readable technologies include magnetic recording, waveform processing, and barcodes. In certain embodiments, the machine-readable device is a solid-state device. Optical character recognition (OCR) can be used to enable machines to read information usable by humans. Any information retrievable by any form of energy can be machine-readable. Furthermore, any data stored on a machine-readable medium can be transferred by streaming over a network. In certain embodiments, the machine-readable medium is a network server disk, e.g., an Internet server disk, e.g., a disk array. In certain embodiments, the machine-readable medium is a network of two or more The term "orthomosaic image" is art-recognized and is used herein to describe a composite image that includes some or all of the images captured by a UAV. These images are "fused" together to generate an orthomosaic image, and the spaces within the image are edited to represent real distances. A module or technique generates the orthomosaic image. An example of a known module capable of generating an orthomosaic image is Pix4Dmapper. In certain embodiments, the module that generates the orthomosaic image can also generate an associated world file that contains data corresponding to the location and scale of the orthomosaic image in real space.

[0019] The term "user" is used herein to describe any person who interfaces with the inventive tools described herein using electronic means, such as a computer or mobile device. Such users may be licensed or unlicensed and may be given certain access rights in the interface based on such status.

[0020] The term "user interface" is used herein to describe, for example, a graphical user interface (GUI) that allows a user to interface with an application programming interface (API) and input data using interface components such as buttons, text fields, checkboxes, etc.

[0021] II. Explosives Detection and Identification of the Present Invention The present invention supports humanitarian mine action (HMA) operations using UAVs equipped with miniaturized optical, geophysical, visual, and thermal sensors to detect and locate explosives or LUXO, thereby contributing to the field by providing rapid, low-cost data acquisition over large areas and reducing the costs, risks, and time associated with surveying contaminated areas. The resulting increased generation and accessibility of data from UAVs, along with the descriptive ability to analyze these large datasets, avoids data processing delays. In contrast to manual analysis, which can be time-consuming, subjective, and inconsistent, the system and method of the present invention uses computers and machine learning to process datasets to assist in identifying the presence of mines in UAV surveys, enabling stakeholders to more intelligently plan HMA operations. This method will also help reduce the size of search areas in contaminated areas and provide important information regarding which areas should be prioritized for clearance efforts.

[0022] A. Explosives Detection and Identification Method of the Present Invention Accordingly, one embodiment of the present invention is a method for detecting and identifying explosives in a target area using an unmanned aerial vehicle (UAV), the method comprising: storing the image data received from the UAV in a machine-readable medium; processing the image data to generate an image of a training area having known explosives; fusing images of the training area, for example using photogrammetry, to generate an orthomosaic image representing the training area with explosives; analyzing one or more segmented images derived from the orthomosaic image using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks (e.g., convolutional neural networks (CNNs)) trained to detect patterns or predefined objects to generate a training model for automating the detection of explosives; identifying and labeling explosives in a training area in the orthomosaic imagery to improve the training model; applying the improved training model to the target area to predict the location of unknown explosives within the target area by comparing the trained orthomosaic image data from the target area with previously generated orthomosaic image data from the training area; Including, A method is provided, the method being suitable for detecting and identifying explosives within a target area using an unmanned aerial vehicle (UAV).

[0023] In certain embodiments of the present invention, the predicted explosive locations are displayed to the user via an output device selected from a portable smart phone or touch screen tablet. In certain embodiments of the present invention, the predicted explosive locations are displayed to the user via a selected output device from the web application.

[0024] 3, an exemplary method is shown for detecting, locating, and / or identifying LUXO on or within a predetermined area 32. The exemplary illustrated method is further described herein below.

[0025] i.Storing image data received from UAV A method for detecting and identifying explosives within a target area using an unmanned aerial vehicle (UAV) includes storing image data received from the UAV on a machine-readable medium.

[0026] In certain embodiments of the present invention, collection of image data by an aircraft (e.g., a UAV) occurs over multiple passes of a training area, e.g., collecting data over multiple passes in multiple conditions. In certain embodiments, data is collected using multiple image collection techniques and / or imaging devices. For example, a UAV 34 may be flown over a training area several times a day or over several days, thereby generating different image data having different lighting and other environmental conditions.

[0027] In certain embodiments of the present invention, as described above in 401-403, the processing system 111 can use image data and datasets of the same area under different lighting and environmental conditions to detect and identify LUXOs, improving efficiency. In a particular example, one orthomosaic image of area 32 is labeled with a rough bounding box label for the orthomosaic image of that area. The locations of ground control points set by the user are used to further refine the location of the bounding box label. However, these bounding box labels are typically slightly offset from the location of the mines in the orthomosaic image. Note that in this case, the orthomosaic image still needs to be processed, as described above in relation to 403, to apply labels and / or adjust the labels (e.g., bounding boxes) to accurately delineate the object of interest.

[0028] In particular embodiments, at 401, UAV 31 is flown near and / or over area 32 to image ground G within area 32 using imaging device 34. In one example, UAV 31 is flown over area 32 that is believed to be or is known to be contaminated with LUXO. The flight path of UAV 31 over area 32 to properly image area 32 may depend on flight conditions (e.g., temperature, wind, precipitation), imaging device 34 specifications (e.g., pixels, zoom), and / or UAV 31 specifications (e.g., airspeed, duration of operation between refueling / charging). The flight path may be communicated to UAV 31 by a user via remote control device 35 and selected by the user based on analysis of image data relayed to control system 100 to meet predetermined image requirements for the area (e.g., percentage overlap between adjacent images). In one non-limiting example, UAV 31 is flown along a predetermined rectangular flight path at an altitude of approximately 8.0 to 12.0 meters above ground G. In this specific example, UAV 31 flies at a speed of 1.0 meter per second and captures images / photographs of a portion of area 32 every 1.50 seconds, such that the overlap between adjacent images is 80.0 to 85.0% of the forward and lateral overlap between transects. The distance between UAV 31 and ground G may vary, and system 30 may also monitor and communicate GPS data as UAV 31 flies along the flight path. Note that UAV 31 may log and transmit GPS data as it flies along the flight path and / or tag each captured image with corresponding GPS data. In a specific example, images are tagged with GPS data based on an internal drone GPS device. In another example, images are tagged with GPS data based on ground control points surrounding area 32. For example, the GPS data may be based on a global navigation satellite system (GNSS), such as the Trimble Zephyr3.

[0029] In certain embodiments of the present invention, the image data collected from the UAV is multispectral data. In certain embodiments of the present invention, the transmission of image data to the control system is wired or wireless.

[0030] ii. processing the image data to generate an image of the training area; A method for detecting and identifying explosives in a target area using an unmanned aerial vehicle (UAV) includes processing image data to generate an image of a training area having known explosives.

[0031] In certain embodiments of the present invention, image data from the imaging device 34 is provided to a processing system 111, which may include one or more image processors that process the data to generate images, compare the image data with previously generated images, identify similarities, differences, and / or patterns in the data or images, and / or detect objects, such as LUXO 33, within the data or images. In certain examples, the processing system 111 utilizes tools to determine characteristics of the object (e.g., LUXO 33) imaged by the imaging device 34 relative to the image data. The tools may include obtaining object coordinates and boundaries. In certain examples, the processing system 111 and / or the control system 100 generally incorporate image processing algorithms, techniques, modules, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks trained to detect patterns or predefined objects. In certain examples, the processing system 111 may include an artificial intelligence system (e.g., IBM's Watson Artificial Intelligence). In a particular example, processing system 111 includes a machine learning model based on an architecture called Faster R-CNN implemented by OpenMMLab in a project called MMDetection. In a particular example, processing system 111 includes using one or more of the following methodologies / tools: TensorFlow, Keras, Python, OpenCV, neural networks, deep learning, and / or computer vision.

[0032] It should be noted that in certain examples, the exemplary method may include conducting a survey of a known area 34 having known LUXO 33 present within the area 34, thereby generating image data similar to the image data described above at 401, and further processing the image data from the known area 34 to generate an orthomosaic image similar to the orthomosaic image generated above at 402. Thus, the orthomosaic image associated with the known area 34 having known LUXO 33 forms a known dataset that serves as a starting point for the processing system 111 to process other additional datasets associated with known areas or new areas 34 having unknown LUXO 33. Thus, the datasets (e.g., a dataset of a known area with known LUXO and a dataset of another area with unknown LUXO) build on each other and subsequently serve to train a processing system (e.g., artificial intelligence, neural networks, computer learning) to detect and identify LUXO in unknown locations as the system 30 is used to locate LUXO-added areas 32.

[0033] iii. Fusing the images of the training area to generate an orthomosaic image. A method for detecting and identifying explosives in a target area using an unmanned aerial vehicle (UAV) includes fusing images of a training area using, for example, photogrammetry, to generate an orthomosaic image representing the training area having explosives.

[0034] In certain embodiments of the present invention, the fused image includes both image data and data from other sources, including GPS and map data. iv. Analyzing segmented images derived from orthomosaic images A method for detecting and identifying explosives in a target area using an unmanned aerial vehicle (UAV) includes analyzing one or more segmented images derived from an orthomosaic image using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or an artificial neural network (e.g., a convolutional neural network (CNN)) trained to detect patterns or predefined objects to generate a training model for automating the detection of explosives.

[0035] In certain embodiments of the invention, the output of this step is multiple segmented orthomosaic images with unique naming conventions and / or updated annotation files so that the position of each orthomosaic image relative to the larger, unsegmented orthomosaic image can be determined.

[0036] In certain embodiments of the present invention, an orthomosaic image is segmented into orthomosaic images of smaller file sizes and / or image sizes, which are then processed by a processing system. For example, the processing system 111 processes the segmented orthomosaic images by inputting the segmented orthomosaic images through a neural network or machine learning model. In certain embodiments, the processing system 111 may process only the maximum allowable image size, which is smaller than the unsegmented orthomosaic image. In certain embodiments, the processing system 111 may include techniques to segment the labeled orthomosaic images as described above in 403 and generate data corresponding to the position of the segmented orthomosaic images relative to the unsegmented orthomosaic image as described above. The processing system 111 in this example may also generate an annotation file for each corresponding segmented orthomosaic image. Note that in certain examples, segmenting an orthomosaic image includes cropping an image with a user-defined percentage of overlap with other adjacent segmented orthomosaic images, such that the segmented orthomosaic image is cropped in one image and complete in the adjacent image. For example, this overlap functionality of the processing system 111 may include smart cropping such that user-defined crop sizes are minimally expanded or contracted to ensure that all cropped images of each orthomosaic image have a uniform size.

[0037] In certain embodiments of the invention, once the data has been partitioned and placed in the correct folder structure, processing system 111 awaits an input command from a user to begin training. Once processing system 111 completes a training session with one or more datasets or image data, a completed training file is output and stored in memory system 112, incorporating a model dataset that has been trained to predict classes of LUXO. This model dataset is then used at 406 to predict the presence and location of UXO in other subsequent files, image data, and / or datasets.

[0038] v. Identifying and labeling explosives within the training area A method for detecting and identifying explosives in a target area using an unmanned aerial vehicle (UAV) includes identifying and labeling explosives in a training area in an orthomosaic image to improve a training model.

[0039] In certain embodiments of the present invention, labeling the orthomosaic image includes marking with indicia selected from the group consisting of boxes, geometric systems, alphanumeric text, graphics, colors, and any combination thereof. In certain embodiments, each explosive detected and identified on the labeled orthomosaic image is assigned a quality grade metric. Note that in certain embodiments, after the orthomosaic image is labeled as described above in 403 using additional modules of the processing system 111, thereby, for example, assigning a quality grade metric (e.g., good, medium, poor) to each bounding box, label box, and / or object detected and identified by the processing system 111. Thus, the dataset can be sorted by the quality grade metric to allow a user and / or the control system 100 to remove or relabel the quality grade metric. The orthomosaic, annotations, and world files can be placed in a private GitHub repository.

[0040] In certain embodiments of the present invention, processing system 111 applies labels to the orthomosaic image generated in 402. Labels of objects identified by processing system 111 are added to the orthomosaic image and may be any suitable indicia, such as boxes, geometric systems, alphanumeric text, graphics, colors, etc.

[0041] In certain embodiments of the present invention, the method further comprises generating an associated world file containing training area data corresponding to the position and scale of the orthomosaic image in real space.

[0042] In certain embodiments of the present invention, the method further includes generating an annotation file containing a list of all labeled explosives in the training area and corresponding munitions types and GPS location data.

[0043] In another example, the processing system 111 runs another script to overlay the predicted locations of LUXOs, their classes and confidence scores onto the orthomosaic image from which the predictions were obtained, creating a highly detailed aerial map of the area that illustrates the type and level of contamination in the area.

[0044] In certain embodiments of the present invention, the processing system 111 includes an interface for a user to manually review all predicted locations of LUXOs. In this example, the processing system 111 generates predicted boxes and labels overlaid on the orthomosaic image, allowing the user to view the predicted boxes and determine whether what is contained within the boxes is truly what the machine predicted. Using the predicted locations of LUXOs, the user can quickly and efficiently mark all predictions that appear to be false alarms, leaving behind a list of scrutinized coordinate predictions of UXOs that can be investigated by Explosive Ordnance Disposal (EOD) teams.

[0045] vi. Applying the improved training model to the target area A method for detecting and identifying explosives within a target area using an unmanned aerial vehicle (UAV) includes applying an improved training model to the target area to predict the location of unknown explosives within the target area by comparing trained orthomosaic image data from the target area with previously generated orthomosaic image data from the training area.

[0046] In certain embodiments of the present invention, the step of applying the improved training model to the target area comprises: storing the target area image data on a machine-readable medium; processing the target area image data to generate an image of the target area having the unknown explosive; fusing the images of the target area to generate an orthomosaic image representing the target area having the unknown explosives; predicting the location of unknown explosives within the target area by analyzing one or more segmented images derived from the orthomosaic imagery using an improved training model using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or an artificial neural network (e.g., a convolutional neural network (CNN)) to compare the trained orthomosaic image data from the target area with previously generated orthomosaic image data from the training area; Identifying and labeling explosives within a target area in the orthomosaic image; In certain embodiments, data obtained from identifying and labeling explosives within a target area is used to further refine training models for detecting and identifying unknown explosives.

[0047] In certain embodiments of the present invention, image data collection by an aerial vehicle (e.g., a UAV) occurs over multiple passes of a training area, e.g., collecting data over multiple passes in multiple conditions. In certain embodiments, data is collected using multiple image collection techniques and / or imaging devices. For example, a UAV 34 may be flown over the target area 32 several times a day or over several days, thereby generating different image data with different lighting and other environmental conditions (see 401).

[0048] In certain embodiments of the present invention, the processing system 111 can use image data and datasets of the same area under different lighting and environmental conditions to locate and identify LUXO, as described above in 401-403, improving efficiency. In a particular example, one orthomosaic image of area 32 is labeled with an approximate bounding box label for the orthomosaic image of that area. The locations of ground control points set by the user are used to further refine the location of the bounding box label. Note that the orthomosaic image still needs to be processed, as described above in relation to 403, to apply labels and / or adjust the labels (e.g., bounding boxes) to accurately delineate the objects of interest.

[0049] In particular embodiments, at 401, UAV 31 is flown near and / or over area 32 to image ground G within area 32 using imaging device 34. In one example, UAV 31 is flown over area 32 that is believed to be or is known to be contaminated with LUXO. The flight path of UAV 31 over area 32 to properly image area 32 may depend on flight conditions (e.g., temperature, wind, precipitation), imaging device 34 specifications (e.g., pixels, zoom), and / or UAV 31 specifications (e.g., airspeed, duration of operation between refueling / charging). The flight path may be communicated to UAV 31 by a user via remote control device 35 and selected by the user based on analysis of image data relayed to control system 100 to meet predetermined image requirements for the area (e.g., percentage overlap between adjacent images). In one non-limiting example, UAV 31 is flown along a predetermined rectangular flight path at an altitude of approximately 8.0 to 12.0 meters above ground G. In this specific example, UAV 31 flies at a speed of 1.0 meter per second and captures images / photographs of a portion of area 32 every 1.50 seconds, such that the overlap between adjacent images is 80.0 to 85.0% of the forward and lateral overlap between transects. The distance between UAV 31 and ground G may vary, and system 30 may monitor and communicate GPS data as it flies along the flight path. Note that UAV 31 may log and transmit GPS data as it flies along the flight path and / or tag each captured image with corresponding GPS data. In a specific example, images are tagged with GPS data based on an internal drone GPS device. In another example, images are tagged with GPS data based on ground control points surrounding area 32. For example, the GPS data may be based on a Global Navigation Satellite System (GNSS), such as a Trimble Zephyr 3.

[0050] In certain embodiments of the present invention, the image data collected from the UAV is multispectral data. In certain embodiments of the present invention, the transmission of image data to the control system is wired or wireless.

[0051] In certain embodiments of the present invention, labeling the orthomosaic image includes marking with indicia selected from the group consisting of boxes, geometric systems, alphanumeric text, graphics, colors, and any combination thereof. In certain embodiments, each explosive detected and identified on the labeled orthomosaic image is assigned a quality grade metric. Note that in certain embodiments, after the orthomosaic image is labeled as described above in 403 using additional modules of the processing system 111, thereby, for example, assigning a quality grade metric (e.g., good, medium, poor) to each bounding box, label box, and / or object detected and identified by the processing system 111. Thus, the dataset can be sorted by the quality grade metric to allow a user and / or the control system 100 to remove or relabel the quality grade metric. The orthomosaic, annotations, and world files can be placed in a private GitHub repository.

[0052] In certain embodiments of the invention, the method further comprises generating an associated world file containing target area data corresponding to the position and scale of the orthomosaic image in real space.

[0053] In certain embodiments of the present invention, the method further comprises generating an annotation file containing a list of all labeled explosives in the target area and corresponding munitions types and GPS location data.

[0054] In another example, the processing system 111 runs another script to overlay the predicted locations of LUXOs, their classes and confidence scores onto the orthomosaic image from which the predictions were obtained, creating a highly detailed aerial map of the area that illustrates the type and level of contamination in the area.

[0055] B. Explosives Detection and Identification Processing Module of the Present Invention The method of the present invention may be utilized and implemented as a processing system module or processing system 111. Accordingly, one embodiment of the present invention is an explosives detection and identification processing module including a control system suitable for receiving and processing image data transmitted from an unmanned aerial vehicle (UAV), the module comprising: storing the image data received from the UAV in a machine-readable medium; processing the image data to generate an image of a training area having known explosives; fusing images of the training area, for example using photogrammetry, to generate an orthomosaic image representing the training area with explosives; analyzing one or more segmented images derived from the orthomosaic image using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks (e.g., convolutional neural networks (CNNs)) trained to detect patterns or predefined objects to generate a training model for automating the detection of explosives; identifying and labeling explosives in a training area in the orthomosaic imagery to improve the training model; applying the improved training model to the target area to predict the location of unknown explosives within the target area by comparing the trained orthomosaic image data from the target area with previously generated orthomosaic image data from the training area; a machine-readable medium having stored thereon instructions for execution by a processor to perform a method comprising: The module is suitable for detecting and identifying explosives in a target area using an unmanned aerial vehicle (UAV).

[0056] In a particular embodiment of the present invention, the control system is adapted to receive and process positioning data from a GPS sensor component, and such data is fused into an orthomosaic image.

[0057] In certain embodiments of the present invention, as described above in 401-403, the processing system 111 can use image data and datasets of the same area under different lighting and environmental conditions to detect and identify LUXO, thereby improving efficiency. In a particular example, a first script (e.g., a list of Python commands executed based on a module of the processing system 111 and / or stored in the memory system 112) is used to label one orthomosaic image of the area 32 with approximate bounding box labels for the orthomosaic image of that area. A second script further refines the location of the bounding box labels using the locations of ground control points set by the user. However, these bounding box labels are typically slightly offset from the location of the mines in the orthomosaic image. Note that in this case, the orthomosaic image still needs to be processed using the processing system 111, as described above in relation to 403, to apply labels and / or adjust the labels (e.g., bounding boxes) to accurately delineate the object of interest.

[0058] In certain embodiments, images of area 32 are captured by a UAV and communicated to the control system (e.g., wirelessly or via a wired connection between the UAV and a port on remote control device 35 that is in communication with control system 100 during or after the flight of UAV 31). In certain examples, processing system 111 processes the image data using one or more methodologies, tools, models, etc. In certain examples, processing system 111 generates an orthophoto or orthomosaic image of the image data generated by the UAV in 402.

[0059] In certain embodiments of the present invention, after the orthomosaic images are segmented, the dataset and / or segmented images are stored in memory system 112 and may include the segmented orthomosaic images along with corresponding annotation files (including labels), world files (including real-world spatial information), and / or one annotation file describing all labeled munitions in those images and a metadata file describing the crop and overlap sizes obtained for each orthomosaic (which is important information when locating coordinates within the segmented images in real space).

[0060] In one particular, non-limiting example, the labels applied to the orthomosaic image are rectangular boxes with alphanumeric text corresponding to the LUXO dataset stored in memory system 112. In particular examples, processing system 111 uses a machine learning model or neural network to determine the LUXO in the orthomosaic image or segmented orthomosaic image. In particular examples, the dataset and the orthomosaic image or segmented orthomosaic image are stored in memory system 112. The applied labels may include the class or name of the corresponding munitions, as described in 401, so that processing system 111 can continue to form and / or refine a general model of the particular munitions as additional areas 34 are surveyed. Thus, in particular examples, processing system 111 (e.g., a computer learning model or neural network) is “trained” or “learns” to detect and label similar munitions in different survey areas 34. The processing system 111 can also generate an annotation file containing a list of all labeled LUXOs and corresponding data (eg, type of munition, such as projectile, grenade, or anti-personnel mine, GPS location data).

[0061] In particular embodiments, once the processing system 111 (e.g., a machine learning model, a neural network) has been trained, the control system 110 includes the trained processing system 111 (e.g., a machine learning model), and labeled data for evaluating the model's performance is processed by the control system 100, or unlabeled image data from new areas 32 for generating predictions of LUXO presence and location is generated at 406. The first step toward these goals is to discover at which epoch the model performed best. Training data is provided to the processing system 111 in stages called epochs. At each epoch, the training data is provided to the processing system 111 memory system 112 in randomized batches to improve variability and efficiency. With each epoch, the model becomes increasingly better at identifying the exact object of interest in the training set, but this may mean that it becomes worse at forming a model that generalizes well enough to identify images in the test set (data on which the model was not trained). When processing system 111 "sees" an object with a particular size, color, and orientation, it is trained only to detect objects with that size, color, and orientation. It is important to train one or more aspects of processing system 111 on a wide variety of training data, but not too many epochs, otherwise it will specialize in identifying images in the training set and will not be able to detect munitions in other data.

[0062] In certain embodiments of the present invention, the processing system 111 is further configured to use another technique, such as a script, to output a graph that makes it very easy to identify the epoch in which the model performed best. The trained model from this epoch is then used to evaluate its performance on a test set to assess how accurate the model is in identifying UXOs in unknown locations. Graphs and spreadsheets are output from the processing system to assess the accuracy of the model. Accuracy is calculated at the orthomosaic image level, which means that the processing system 111 determines the location of the predicted box within each orthomosaic image or images. This provides the user with useful statistics on how many objects were correctly detected, incorrectly detected, and missed in each orthomosaic image.

[0063] In certain embodiments of the present invention, the processing system 111 executes a script for calculating and outputting actual coordinate predictions of the predicted LUXO objects. In another example, the processing system 111 runs another script to overlay the predicted locations of LUXOs, their classes and confidence scores onto the orthomosaic image from which the predictions were obtained, creating a highly detailed aerial map of the area that illustrates the type and level of contamination in the area.

[0064] C. System for Detecting and Identifying Explosives of the Present Invention The methods and processing modules of the present invention may function as components of, and be implemented as, a system that includes additional components such as a UAV, a remote control device, an output device, and / or a GPS sensor component. Accordingly, another embodiment of the present invention is a system for detecting and identifying explosives in a target area using an unmanned aerial vehicle (UAV), comprising: a UAV adapted to include an imaging device suitable for capturing image data and a transmission component capable of transmitting the image data to a control system; 1. An explosives detection and identification processing module including a control system suitable for receiving and processing image data transmitted from a UAV, the module comprising: storing the image data received from the UAV in a machine-readable medium; processing the image data to generate an image of a training area having known explosives; fusing images of the training area, for example using photogrammetry, to generate an orthomosaic image representing the training area with explosives; analyzing one or more segmented images derived from the orthomosaic image using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks (e.g., convolutional neural networks (CNNs)) trained to detect patterns or predefined objects to generate a training model for automating the detection of explosives; identifying and labeling explosives in a training area in the orthomosaic imagery to improve the training model; applying the improved training model to the target area to predict the location of unknown explosives within the target area by comparing the trained orthomosaic image data from the target area with previously generated orthomosaic image data from the training area; an explosives detection and identification processing module including a machine-readable medium having stored thereon instructions for execution by a processor to perform a method including: Including, The system is directed to a system suitable for detecting and identifying explosives within a target area using an unmanned aerial vehicle (UAV).

[0065] In certain embodiments of the present invention, the system further includes a GPS sensor component disposed on the UAV, adapted to capture positioning data and capable of transmitting said image data to a control system, the control system being adapted to receive and process the positioning data from the GPS sensor component, such data being fused into an orthomosaic image.

[0066] In certain embodiments of the present invention, the imaging device is selected from the group consisting of a camera, a visible light sensor, a multispectral sensor, a thermal sensor, and any combination thereof.

[0067] In certain embodiments, UAV 31 communicates with and is controlled by remote control device 35. FIG. 1 depicts remote control device 35 as a stationary unit located on ground G, which can be operated by a user. In other examples, remote control device 35 is a hand-carried module that can be carried by a user or a ground vehicle. In yet other examples, remote control device 35 is incorporated into or attached to a mobile ground vehicle (e.g., a tank, a Humvee). Remote control device 35 and / or UAV 31 may be in communication with and / or part of control system 100 (described herein, see FIG. 2 ), thereby transferring data between UAV 31, remote control device 35, and / or control system 100. The remote control device 35 and / or UAV 31 may also communicate with a mobile data / telephone network and / or satellites 36 so that data can be transferred to different components of the system 30 and global positioning data (e.g., GPS coordinates) can be determined.

[0068] In certain embodiments of the present invention, control module 100 communicates with each of one or more components of system 30 via communication link 110, which may be any wired or wireless link. Control module 100 can control one or more operational characteristics of system 30 and its various subsystems by receiving information and / or sending and receiving control signals via communication link 110. In one example, communication link 100 is a controller area network (CAN) bus, although other types of links may be used. It will be appreciated that the scope of connections and communication link 110 may actually be one or more shared connections or links between some or all of the components in system 30. Furthermore, the lines of communication link 110 are intended only to indicate that the various control elements can communicate with each other, and do not represent actual hardwired connections between the various elements or the only path of communication between the elements. Additionally, system 30 can incorporate various types of communication devices and systems; thus, the illustrated communication link 110 may actually represent various different types of wireless and / or wired data communication systems.

[0069] In certain embodiments of the present invention, control system 100 may be a computing system including a processing system 111, a memory system 112, and an input / output (I / O) system 113 for communicating with other devices, such as input devices 120 (such as imaging device 34 or GPS sensors / devices) and output devices 130 (either of which may additionally or alternatively be stored in cloud 140). Processing system 111 loads and executes executable programs 114 from memory system 112, accesses data 115 stored within memory system 112, and directs system 30 to operate as described in further detail herein.

[0070] In certain embodiments of the present invention, machine-readable media or memory system 112 may include any storage medium readable by processing system 111 and capable of storing executable programs 114 and / or data 115. Memory system 112 may be implemented as a single storage device or may be distributed across multiple storage devices or subsystems that cooperate to store computer-readable instructions, data structures, program modules, or other data. Memory system 112 may include volatile and / or nonvolatile systems and may include removable and / or non-removable media implemented in any method or technology for storing information. Storage media may include non-transitory storage media, including, for example, random access memory, read-only memory, magnetic disks, optical disks, flash memory, virtual and non-virtual memory, magnetic storage devices, or any other medium that can be used to store information and that can be accessed by an instruction execution system.

[0071] In particular embodiments of the present invention, processing system 111 may be implemented as a single microprocessor or other circuit, or may be distributed across multiple processing devices or subsystems that cooperate to execute executable programs 114 from memory system 112. Non-limiting examples of processing systems include general-purpose central processing units, application-specific processors, and logic devices.

[0072] In certain embodiments of the present invention, the predicted LUXO locations are displayed to a user via an output device 130, such as a mobile smartphone, touchscreen tablet, or web application. In certain embodiments, the processing system 111 executes scripts to calculate and output actual coordinate predictions of the predicted LUXO objects. These coordinates can be easily viewed in any geographic information system (GIS), such as Google Earth Pro or QGIS. Along with each predicted object location is a predicted label of which class of munition it belongs to and a confidence score that represents how confident the machine is that there is a mine at this location.

[0073] Certain aspects of the present disclosure are described or depicted as functional and / or logical block components or processing steps, which may be performed by any number of hardware, software, and / or firmware components configured to perform the specified functions. For example, certain embodiments employ integrated circuit components, such as memory elements, digital signal processing elements, logic elements, look-up tables, etc., configured to perform various functions under the control of one or more processors or other control devices. Connections between functional and logical block components are exemplary only and may be direct or indirect and may follow alternate paths. [Example]

[0074] Example Example 1 Exemplary embodiment: System for detecting and identifying explosives Exemplary embodiments of the system of the present invention are described herein below. Referring to Figure 1, an exemplary system 30 for detecting and locating land mines and other unexploded ordnance (UXO) (hereinafter collectively referred to as "LUXO") is shown. As described further herein below, the present disclosure includes methods for automating the detection of LUXO on the ground from unpiloted aerial vehicle-based (UAV) imagery using convolutional neural networks (CNNs) and / or machine learning models.

[0075] The system 30 generally includes a UAV 31 flown near and / or over a field or area of ​​interest 32 while the LUXO 33 is on or embedded in the ground G. The UAV 31 may be any suitable aircraft capable of flying near or over the area 32 and imaging the area 32 (as further described herein). In a particular example, the UAV 31 is a commercially available quadcopter or hexacopter adapted for the purposes of the present invention, equipped with an imaging device 34 such as a camera, visible light sensor, multispectral sensor, and / or thermal sensor. In a specific example, the UAV is an adapted aerial drone such as a DJI Matrice 600, a DJI Phantom 4, or a DJI Mavic 2 Enterprise Dual.

[0076] Referring to the image on the right of FIG. 5 , an example image of an area 32 having LUXO 33 captured by the imaging device 34 of the UAV 31 is shown. The image on the left of FIG. 5 is an example processed image processed and labeled by the processing system 111. The image on the bottom of FIG. 5 shows a graphical depiction of sample output data from the processing system 111, including labeling an orthomosaic image used to train the processing system 111, detecting and identifying LUXO in other images using the trained processing system 111, and predicted object coordinates and grade scores. Referring to the three left images of FIG. 6 , the processing system 111 generates a labeled orthomosaic image based on image data from the imaging device 34. The processing system 111 detects and labels LUXO and further generates segmented orthomosaic images of the labeled orthomosaic image. One or more segmented orthomosaic images are used to train a machine learning model or neural network. Thus, referring to the image 4 on the right, FIG. 6, the processing system 111 includes a trained machine learning model or neural network that can process the unlabeled orthomosaic images, segment the unlabeled orthomosaic images to generate overlapping segmented unlabeled orthomosaic images, and use the trained machine learning model or neural network to detect and identify LUXOs in the segmented unlabeled orthomosaic images. The processing system can further output data for each identified LUXO and generate one or more GIS shapefiles. FIG. 7 shows two example orthomosaic images with dimensions and one example orthomosaic image with overlapping orthomosaic images.

[0077] Incorporation by Reference The entire contents of all patents, published patent applications, and other publications cited herein are expressly incorporated herein by reference in their entirety.

[0078] equivalent Specific terminology has been used herein for brevity, clarity, and understanding. Such terminology is used for descriptive purposes and is intended to be broadly construed, so that no unnecessary limitations are to be inferred therefrom. The different devices, systems, and method steps described herein may be used alone or in combination with other devices, systems, and methods. It is understood that various equivalents, alternatives, and modifications are possible within the scope of the appended claims.

[0079] Those skilled in the art will recognize, or be able to ascertain using no more than routine experimentation, numerous equivalents to the specific procedures described herein. Such equivalents are considered to be within the scope of this invention and are covered by the following claims. Furthermore, any numerical or alphabetical ranges provided herein are intended to include both the upper and lower limits of the range. In addition, any listing or grouping is intended, in at least one embodiment, to represent a shorthand or convenient way of listing independent embodiments, and therefore each item in the list should be considered a separate embodiment.

Claims

1. 1. A system for detecting and identifying explosives within a target area using an unmanned aerial vehicle (UAV), comprising: a UAV adapted to include an imaging device suitable for capturing image data and a transmission component capable of transmitting the image data to a control system; an explosives detection and identification processing module including the control system adapted to receive and process the image data transmitted from the UAV, the module comprising: storing the image data received from the UAV in a machine-readable medium; processing the image data to generate an image of a training area having known explosives; fusing the images of the training area to generate an orthomosaic image representing the training area with the explosives; analyzing one or more segmented images derived from said orthomosaic imagery using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks trained to detect patterns or predefined objects to generate a training model for automating said detection of explosives; identifying and labeling the explosives in the training area in the orthomosaic image to improve the training model; applying the improved training model to a target area to predict the location of unknown explosives within the target area by comparing trained orthomosaic image data from the target area with previously generated orthomosaic image data from a training area; an explosives detection and identification processing module including said machine-readable medium having stored thereon instructions for execution by a processor to perform a method comprising: Including, The system is suitable for detecting and identifying explosives within the target area using the unmanned aerial vehicle (UAV).

2. 2. The system of claim 1, further comprising a GPS sensor component disposed on the UAV, adapted to capture positioning data and capable of transmitting the image data to the control system, the control system adapted to receive and process the positioning data from the GPS sensor component, the data being fused into the orthomosaic image.

3. applying the improved training model to the target area; storing the target area image data on the machine-readable medium; processing the target area image data to generate an image of the target area having unknown explosives; fusing the images of the target area to generate an orthomosaic image representing the target area having the unknown explosive; predicting the location of unknown explosives within the target area by analyzing one or more segmented images derived from the orthomosaic imagery using the improved training model using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks (e.g., convolutional neural networks (CNN)) to compare the trained orthomosaic image data from the target area with previously generated orthomosaic image data from a training area; identifying and labeling the explosives within the target area in the orthomosaic image; The system of claim 1 , comprising:

4. 4. The system of claim 3, wherein the data obtained in the identification and labeling of the explosives in the target area is used to further refine the training model for detecting and identifying unknown explosives.

5. The system of claim 1 , wherein the imaging device is selected from the group consisting of a camera, a visible light sensor, a multispectral sensor, a thermal sensor, and any combination thereof.

6. 4. The system of claim 3, wherein the step of labeling the orthomosaic image comprises marking with indicia selected from the group consisting of boxes, geometric systems, alphanumeric text, graphics, colors, and any combination thereof.

7. 10. The system of claim 1, wherein the method further comprises generating an annotation file containing a list of all labeled explosives and corresponding munitions types and GPS location data.

8. 10. The system of claim 1, wherein the predicted location of the explosives is displayed to the user via a selected output device from a web application.

9. 1. An explosives detection and identification processing module including the control system adapted to receive and process the image data transmitted from an unmanned aerial vehicle (UAV), the module comprising: storing the image data received from the UAV in a machine-readable medium; processing the image data to generate an image of a training area having known explosives; fusing the images of the training area to generate an orthomosaic image representing the training area with the explosives; analyzing one or more segmented images derived from said orthomosaic imagery using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks trained to detect patterns or predefined objects to generate a training model for automating said detection of explosives; identifying and labeling the explosives in the training area in the orthomosaic image to improve the training model; applying the improved training model to a target area to predict the location of unknown explosives within the target area by comparing trained orthomosaic image data from the target area with previously generated orthomosaic image data from a training area; said machine-readable medium having stored thereon instructions for execution by a processor to perform a method comprising: An explosives detection and identification processing module, said module being suitable for detecting and identifying explosives within said target area using said unmanned aerial vehicle (UAV).

10. 10. The explosives detection and identification processing module of claim 9, wherein the control system is adapted to receive and process positioning data from a GPS sensor component, the data being fused into the orthomosaic image.

11. applying the improved training model to the target area; storing the target area image data on the machine-readable medium; processing the target area image data to generate an image of the target area having unknown explosives; fusing the images of the target area to generate an orthomosaic image representing the target area having the unknown explosive; predicting the location of unknown explosives within the target area by analyzing one or more segmented images derived from the orthomosaic imagery using the improved training model using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks to compare the trained orthomosaic image data from the target area with previously generated orthomosaic image data from a training area; identifying and labeling the explosives within the target area in the orthomosaic image; 10. The explosives detection and identification processing module of claim 9, comprising:

12. 12. The explosives detection and identification processing module of claim 11, wherein the data obtained in the identification and labeling of the explosives in the target area is used to further refine the training model for detecting and identifying unknown explosives.

13. 12. The explosives detection and identification processing module of claim 11, wherein the step of labeling the orthomosaic image comprises marking with indicia selected from the group consisting of boxes, geometric systems, alphanumeric text, graphics, colors, and any combination thereof.

14. 10. The explosives detection and identification processing module of claim 9, wherein the method further comprises generating an annotation file containing a list of all labeled explosives and corresponding munitions type and GPS location data.

15. 10. The explosives detection and identification processing module of claim 9, wherein the predicted explosives locations are displayed to the user via a selected output device from a web application.

16. 1. A method for detecting and identifying explosives in a target area using an unmanned aerial vehicle (UAV), the method comprising: storing the image data received from the UAV in a machine-readable medium; processing the image data to generate an image of a training area having known explosives; fusing the images of the training area to generate an orthomosaic image representing the training area with the explosives; analyzing one or more segmented images derived from said orthomosaic imagery using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks trained to detect patterns or predefined objects to generate a training model for automating said detection of explosives; identifying and labeling the explosives in the training area in the orthomosaic image to improve the training model; applying the improved training model to a target area to predict the location of unknown explosives within the target area by comparing trained orthomosaic image data from the target area with previously generated orthomosaic image data from a training area; Including, A method for detecting and identifying explosives, wherein the method is suitable for detecting and identifying explosives in the target area using the unmanned aerial vehicle (UAV).

17. applying the improved training model to the target area; storing the target area image data on the machine-readable medium; processing the target area image data to generate an image of the target area having unknown explosives; fusing the images of the target area to generate an orthomosaic image representing the target area having the unknown explosive; predicting the location of unknown explosives within the target area by analyzing one or more segmented images derived from the orthomosaic imagery using the improved training model using image processing algorithms, computer machine learning, computer vision machine learning (CVML), and / or artificial neural networks (e.g., convolutional neural networks (CNN)) to compare the trained orthomosaic image data from the target area with previously generated orthomosaic image data from a training area; identifying and labeling the explosives within the target area in the orthomosaic image; 17. A method for detecting and identifying explosives as set forth in claim 16, comprising:

18. 20. The method for detecting and identifying explosives of claim 17, wherein the data obtained in the identification and labeling of the explosives in the target area is used to further refine the training model for detecting and identifying unknown explosives.

19. 20. The method for detecting and identifying explosives of claim 17, wherein the step of labeling the orthomosaic image comprises marking with indicia selected from the group consisting of boxes, geometric systems, alphanumeric text, graphics, colors, and any combination thereof.

20. 20. The method for detecting and identifying explosives of claim 17, wherein said method further comprises the step of generating an annotation file containing a list of all labeled explosives and corresponding munition type and GPS location data.

21. 20. The method for detecting and identifying explosives of claim 17, wherein the predicted location of the explosive is displayed to the user via a selected output device from a web application.

Images