Abstract: In a particular implementation, a codebook C can be used for quantizing a feature vector of a database image into a quantization index, and then a different codebook (B) can be used to approximate the feature vector based on the quantization index. The codebooks B and C can have different sizes. Before performing image search, a lookup table can be built offline to include distances between the feature vector for a query image and codevectors in codebook B to speed up the image search. Using triplet constraints wherein a first image and a second image are indicated as a matching pair and the first image and a third image as non-matching, the codebooks B and C can be trained for the task of image search. The present principles can be applied to regular vector quantization, product quantization, and residual quantization.

Abstract: An image (e.g., a picture) is received from a communication device. The image has associated metadata that is captured while the image is taken. For example, the metadata may be a location, a date, and a time of the picture. One or more object(s) (e.g., a car) along with characteristics of the object(s) in the image are identified. For example, a characteristic may be an action associated with the object, such as a person is running. In response to analyzing the one or more objects for the one or more characteristics in conjunction with the metadata, one or more potential workflows are determined (e.g., a workflow to call the police). A list that identifies the one or more potential workflows is sent to the communication device. Input to select one of the one or more workflows is received, which initiates the selected workflow.

Abstract: This invention utilizes the random unit to unit physical differences, characteristics, inherent in any manufacturing process, especially such processes which include selected key marking or printing elements of the articles design. The key elements selected are those which facilitate detection of the unit to unit differences generated, in an uncontrolled manner, by the manufacturing process. The visual appearance to the naked eye of the unit to unit differences of such key elements usually falls below the threshold of human perception, but enables their use for ant-counterfeiting purposes.

Abstract: Provided are mechanisms and processes for visual feature tagging in multi-view interactive digital media representations (MIDMRs). In one example, a process includes receiving a visual feature tagging request that includes an MIDMR of an object to be searched, where the MIDMR includes spatial information, scale information, and different viewpoint images of the object. A visual feature in the MIDMR is identified, and visual feature correspondence information is created that links information identifying the visual feature with locations in the viewpoint images. At least one image associated with the MIDMR is transmitted in response to the feature tagging request.

Abstract: An image monitoring system includes: recording means for recording an image captured by a camera via a network; control means for controlling the system so as to display the present image captured by the camera or a past image recorded on the recording means on display means; and moving-object detecting means for detecting a moving object from the image captured by the camera; wherein the moving-object detecting means includes resolution conversion means for generating an image with a resolution lower than the resolution of the image captured by the camera, positional-information output means for detecting a moving object from the image generated by the resolution conversion means and outputting positional information on the detected moving object, and information merging means for merging the positional information of the moving object with the image captured by the camera on the basis of the positional information of the moving object output by the positional-information output means.

Abstract: A computer implemented recognition process of an object in a query image provides a set of training images, each training image being defined by a plurality of pixels and comprising an object tag; determines for each training image of the set a plurality of first descriptors, each first descriptor being a vector that represents pixel properties in a corresponding subregion of the associated training image; and selects among the first descriptors a group of exemplar descriptors describing the set of training images, wherein selecting the exemplar descriptors includes determining the first descriptors having a number of repetitions in the set of training images higher than a certain value.

Abstract: In one embodiment, a computing system may access one or more frames of media content with each frame corresponding to at least part of an image. The computing system may automatically detect at least one object in at least one of the one or more frames. The computing system may automatically identify the at least one object. For each of the identified objects, the computing system may automatically determine an annotation corresponding to that object and each annotation may include one or more graphical elements. For each annotation, the computing system may automatically associate that annotation for display with its corresponding object.

Abstract: Methods and systems for automatically tracking and analyzing imagery data of at least one vehicle on a racetrack comprising. A video event management system with a plurality of video cameras positioned around a racetrack determines the presence of the at least one vehicle and based on a weighted event score corresponding to dynamics for the at least one vehicle and other objects captures video imagery and stills and generates at least one subframe. Excess video imagery data and excess stills data are discarded based on metadata of linked subframes.

Abstract: Techniques and systems are provided for maintaining blob trackers for one or more video frames. For example, a blob tracker is identified for a current video frame. The blob tracker is associated with a blob detected for the current video frame. The blob includes pixels of at least a portion of one or more objects in the current video frame. A current characteristic of pixels in a region of the current video frame associated with the blob tracker is determined. A previous characteristic of pixels in a region of a previous video frame associated with the blob tracker is also determined. A difference is determined between the current characteristic and the previous characteristic, and a status of the blob tracker is determined based on the determined difference. The status of the blob tracker indicating whether to maintain the blob tracker for the one or more video frames.

Abstract: A surveillance system may comprise one or more computing devices and a plurality of robotic surveillance devices. The one or more computing devices may be configured to obtain video data captured by one or more cameras. The one or more computing devices may analyze the video data to identify two or more conditions, and associate results of the identification with respect to the two or more conditions to determine an activity category. The one or more computing device may assign tasks to the plurality of robotic surveillance devices based on the determined activity category. The plurality of robotic surveillance devices may be configured to perform responding actions in response to receiving the tasks.

Abstract: A dynamic signal to noise ratio tracking system enables detection of vehicles within the field of view of the tracking system. The tracking system may include an emitter configured to emit electromagnetic radiation within an area, a detector configured to detect electromagnetic radiation reflected back from vehicles within the area, and a control unit configured to evaluate signals from the detector and control various automated devices as a result of this evaluation.

Abstract: A camera assembly, including: an enclosure having at least two mounting surfaces that are orthogonal to each other for alignment with at least two orthogonal surfaces against which the camera assembly is to be mounted; at least one imaging apparatus disposed within the enclosure and having a predetermined orientation with respect to the enclosure; and a communication device disposed within the enclosure; and a server disposed at a location remote from where the camera is mounted. The camera assembly and the server communicate over a computer communication network to identify at least one mounting measurement of the camera assembly to establish a mapping from an image coordinate system for images generated by the imaging apparatus and a real world coordinate system aligned with an orientation defined by the at least two orthogonal surfaces.

Abstract: A camera assembly, including: an enclosure having at least two mounting surfaces that are orthogonal to each other for alignment with at least two orthogonal surfaces against which the camera assembly is to be mounted; at least one imaging apparatus disposed within the enclosure and having a predetermined orientation with respect to the enclosure; and a communication device disposed within the enclosure; and a server disposed at a location remote from where the camera is mounted. The camera assembly and the server communicate over a computer communication network to identify at least one mounting measurement of the camera assembly to establish a mapping from an image coordinate system for images generated by the imaging apparatus and a real world coordinate system aligned with an orientation defined by the at least two orthogonal surfaces.

Abstract: A camera assembly, including: an enclosure having at least two mounting surfaces that are orthogonal to each other for alignment with at least two orthogonal surfaces against which the camera assembly is to be mounted; at least one imaging apparatus disposed within the enclosure and having a predetermined orientation with respect to the enclosure; and a communication device disposed within the enclosure; and a server disposed at a location remote from where the camera is mounted. The camera assembly and the server communicate over a computer communication network to identify at least one mounting measurement of the camera assembly to establish a mapping from an image coordinate system for images generated by the imaging apparatus and a real world coordinate system aligned with an orientation defined by the at least two orthogonal surfaces.

Abstract: A method and device for the automated driving of a motorized transportation vehicle having at least one camera and an evaluation unit for evaluating the data captured by the camera, wherein lane markings are captured by the camera and the evaluation unit determines traffic lanes from the captured data of the camera, wherein the evaluation unit observes changes in the lane markings which will cause a new traffic lane to be determined over a predefined distance, wherein the new traffic lane is adopted only in response to the changed lane marking being captured over the predefined distance.

Abstract: The present disclosure discloses a method and apparatus for acquiring information. A specific embodiment of the method includes: collecting characteristic information of a road object, the characteristic information comprising: a position, in a world coordinate system, of laser points corresponding to the road object in a laser point cloud collected in an area where the road object is located, and a position in a reflectance map corresponding to a position of a control point of the road object in the world coordinate system; and determining whether a preset accuracy is acceptable based on a match between the characteristic information of the road object and a preset characteristic information of the road object, the preset accuracy comprising: a splicing accuracy of the spliced laser point cloud, an accuracy of the reflectance map, and an accuracy of the high-precision map.

Abstract: A method and system is provided for generating a perception scene graph (PSG) for use in a motor vehicle. External sensor information and vehicle-to-everything (V2X) information are collected on a volume of space surrounding the vehicle. The sensor information and V2X information are fused to increase the fidelity of the information. The fused information is analyzed to detect, classify, and locate objects present in the volume of space surrounding the vehicle. A perception controller generates and publishes the PSG, which includes a virtual 3-D model of the volume of space containing the classified and located objects. At least one vehicle controller subscribes to the PSG and extracts information from the PSG for use as an input for a software application to generate instructions for use in execution of a vehicle application routine. The vehicle controller may also publish a real time copy of the PSG.

Abstract: A method and system is provided for using a scene detection schema for classifying objects in a perception scene graph for a motor vehicle. The scene detection schema includes collecting sensor information about an area surrounding the vehicle, including capturing an image of the area; processing the sensor information to generate a perception scene graph (PSG); detecting a plurality of objects by analyzing the captured image; comparing the detected objects with reference objects; classifying the detected objects based on matching reference objects; and assigning a priority to each of the classified objects. The steps of detecting and classifying objects having higher priorities are reiterated with a greater frequency than objects having lower priorities. An object is assigned a higher priority level if the object is in a focus region of the PSG. The fidelity of the focus region is increased by updating the classification of the objects after each reiteration.

Abstract: A method of distinguishing a three dimensional object from a two dimensional object using a vehicular system includes acquiring image frames captured by a vehicle camera while the vehicle is in motion. First and second feature points are selected from a first detected object in a first captured image frame and tracked in at least a second captured image frame. Third and fourth feature points are selected from a second detected object in the first captured image frame and tracked over at least the second captured image frame. Movements of the first and second feature points over the multiple captured image frames are compared to movements of the third and fourth feature points the multiple captured image frames to distinguish the first object as a three dimensional object and the second object as a two dimensional object.

Abstract: A method of recognizing a biological feature is provided. In an example, the method includes: first biological feature data is obtained; a first recognition operation is performed according to the first biological feature data and biological feature template data to obtain a first recognition result; when the first recognition result indicates a match failure, second biological feature data is obtained; and a re-recognition operation is performed according to the second biological feature data and the biological feature template data to obtain a second recognition result. The second biological feature data and the first biological feature data are collected by a same biological feature collector at different moments in a same biological feature recognition process.

Abstract: In an automatic sampling device for automatically collecting a sample from a sample-holding container having a top surface on which a plurality of sample-holding portions as well as one or more reference points for position adjustment whose relative position to the sample-holding portions is previously known are provided, the automatic sampling device includes: a housing having an opening on a front side for allowing the sample-holding container to be set inside, the housing configured so that a closed space is formed inside when the opening is closed; a sample collector for collecting a sample from any one of the sample-holding portions of the sample-holding container set within the housing; and an appearance information acquisition assistant which assists in an acquisition of appearance information including the sample collector and the one or more reference points for position adjustment of the sample-holding container set within the housing.

Abstract: A supplemental device for attachment to an injection device, the supplemental device comprising: an imaging arrangement configured to capture an image of one or more numbers present on a moveable component of the injection device; and a processor arrangement configured to run a first algorithm to: receive data representing an image; determine the position of at least one number in the image by determining the boundaries of each visible number; select one of the visible numbers to be identified; perform a pattern matching process to identify the value of the selected number; and determine a dose dialled into the injection device using the determined position of the selected one of the at least one visible number in the image.

Abstract: A security camera system capable of reducing an amount of information without degrading useful information is provided. The security camera system includes a pixel array, an image signal processor that performs various types of image processing on raw data including each pixel value from the pixel array, and a pre-stage processing unit arranged between the pixel array and the image signal processor. The pre-stage processing unit includes an important area extraction circuit that receives raw data from the pixel array with respect to each frame and divides an effective pixel area of the frame into an important area and an unimportant area, and transmits the raw data of the important area to the image signal processor.

Abstract: A shape recognition device that recognizes a shape of an object having an indefinite shape and flexibility, and assembled by a robot, the shape recognition device including: an imaging unit that images the object; an image processing unit that recognizes the shape of the object on the basis of the object imaged by the imaging unit; and a simulation processing unit that simulates the shape of the object on the basis of the image of the object imaged by the imaging unit. The simulation processing unit interpolates a recognition result of the shape of the object by the image processing unit, on the basis of a simulation result of the shape of the object.

Abstract: Provided is an image classification system for classifying an input image including: a network constructed so that feature amount extraction processing and fully connected processing are executed on input image data; a path extracting means configured to extract a processing path in the network; and a determination module configured to compare the processing path extracted by the path extracting means and a plurality of setting paths set in advance in association with classes to set a class associated with a matched path as a class of the input image.

Abstract: A method and device for extracting images from PDF documents are disclosed. The method includes performing a text recognition process on a PDF document that includes one or more images. The text recognition process replaces the one or more images with a plurality of contiguous newlines. The method further includes storing a location of each of the one or more images within the PDF document based on occurrence of the plurality of contiguous newlines within the PDF document. The method includes converting each page of the PDF document to an image format in order to generate an image document corresponding to the PDF document. The method further includes extracting each of the one or more images from the image document based on the location stored for each of the one or more images within the PDF document.

Abstract: In a template design application, a user creates a template for a machine-readable image target that supports encoded data and augmented reality homography. The design application enforces design constraints that ensure reliable detection of the target by a detection application while supporting creative freedom in designing the template. The design constraints can include specification of a detection contour outlining a transition between a perimeter of a border shape and an adjacent area of a contrasting color. Data to be encoded is applied to the template to create printable image instances of the target. The template is also loaded into a detection application, which scans and recognizes printed instances of the target and decodes the encoded data. The detection application uses the decoded data to create a synthesized image target matching the scanned target and which is used to execute further tracking of the target within a moving field of view.

Abstract: Described is a system for converting a convolutional neural network (CNN) designed and trained for color (RGB) images to one that works on infrared (IR) or grayscale images. The converted CNN comprises a series of convolution layers of neurons arranged in a set kernels having corresponding depth slices. The converted CNN is used for performing object detection. A mechanical component of an autonomous device is controlled based on the object detection.

Abstract: A medical scan comparison system is operable to receive a medical scan via a network and to generate similar scan data. The similar scan data includes a subset of medical scans from a medical scan database and is generated by performing an abnormality similarity function to determine that a set of abnormalities included in the subset of medical scans compare favorably to an abnormality identified in the medical scan. At least one cross-sectional image is selected from each medical scan of the subset of medical scans for display on a display device associated with a user of the medical scan comparison system in conjunction with the medical scan.

Abstract: One embodiment provides a method, including: receiving, at an information handling device, user input, wherein the user input comprises an indication to perform a function associated with content comprising at least one image; analyzing, using a processor, the at least one image to determine at least one characteristic associated with the at least one image; and providing, based on the analyzing, output describing the at least one characteristic of the at least one image. Other aspects are described and claimed.

Abstract: Some embodiments provide methods for providing images of a person generated by two or more A/V recording and communication devices to one or more users, via a user's client device. For example, first image data may be received from a first A/V recording and communication device at a first location and second image data may be received from a second A/V recording and communication device at a second location. The first image data and the second image data may be analyzed to determine a person depicted in the first image data and a person depicted in the second image data is the same person. In response, a user alert may be generated including data representative of a first facial image of the person and a second facial image of the person. The user alert may then be transmitted to a user's client device.

Abstract: A system includes one or more processors and a memory that stores a generative adversarial network (GAN). The one or more processors are configured to receive a low resolution point cloud comprising a set of three-dimensional (3D) data points that represents an object. A generator of the GAN is configured to generate a first set of generated data points based at least in part on one or more characteristics of the data points in the low resolution point cloud, and to interpolate the generated data points into the low resolution point cloud to produce a super-resolved point cloud that represents the object and has a greater resolution than the low resolution point cloud. The one or more processors are further configured to analyze the super-resolved point cloud for detecting one or more of an identity of the object or damage to the object.

Abstract: Various embodiments can include artificial learning counter surveillance (ALCS) or self-protection surveillance systems (SPSS) and related methods. Apparatuses and methods can include non-destructive electro-optic interference or protection systems as well non-destructive directed energy systems, a control system, and an analysis system for determining an optical system of interest and generating degradation or disruption effects using various equipment items and machine learning systems. Additionally, methods are also provided for determining degree of severity of degradation or disruption based on threshold definitions related to ability to use the optical system of interest for one or more specified applications.

Abstract: In some embodiments, the present invention provides for an exemplary inventive system, including: a communication pipeline, including: at a first end of the communication pipeline: a first processor configured to: obtain a plurality of original content data units having a representative content associated with a subject; apply a trained artificial intelligence algorithm to identify: the representative content of the subject and original background content that is not associated with the subject; remove the original background content to reduce a volume of data being transmitted resulting in an increased capacity of the communication channel; encode and transmit each respective modified content data unit from the first end of the communication pipeline to a second end; a second processor configured to: receive and decode each respective modified content data unit; generate a respective artificial background content; and combine the representative content associated with the subject and the respective artificia

Abstract: Techniques are provided for performing sentiment analysis on words in a first data set. An example embodiment includes generating a word embedding model including a first plurality of features. A value indicating sentiment for the words in the first data set can be determined using a convolutional neural network (CNN). A second plurality of features are generated based on bigrams identified in the data set. The bigrams can be generated using a co-occurrence graph. The model is updated to include the second plurality of features, and sentiment analysis can be performed on a second data set using the updated model.

Abstract: A processor causes a transmission circuit to output a carrier signal intermittently. Each time the carrier signal is output, the processor determines whether or not a level of a detection signal is in a satisfactory state. In a first case where the level of the detection signal is determined to be in the satisfactory state, the processor performs a read control to cause the transmission circuit to perform a precharge process and a transmission process. In a second case where the level of the detection signal is not determined to be in the satisfactory state, the processor determines that the unit attachment portion is in a unit non-attachment state. In the first case, when the processor succeeds in identifying the developer information, the processor determines that the unit attachment portion is in a unit attachment state, and otherwise, determines that the unit attachment portion is in an exceptional state.

Abstract: A mobile terminal that is capable of reducing time and effort of a user when the user gives an execution instruction to an image processing apparatus. A memory device stores a set of instructions. At least one processor executes the set of instructions to detect an image processing apparatus that performs wireless communication, display, on a display unit, a lock screen that accepts an unlocking operation by a user, display, on the display unit, a screen of a control application for controlling an image processing apparatus, after the unlocking operation is accepted, accept, while displaying the lock screen, an execution instruction for controlling the detected image processing apparatus, without accepting the unlocking operation by a user.

Abstract: An item security system and method of verifying items selected for purchase at a checkout station which focus on reducing losses at the checkout station as opposed to exits. One example method includes reading wireless tags on items at a checkout station by a wireless tag reader, reading barcodes on a number of the items at the checkout station by a barcode reader, comparing the number of items with read barcodes to the items with read wireless tags, and issuing an alert when the number of items is less than all of the items.

Abstract: A two-dimensional code initialization method includes: receiving information about any two-dimensional code, and determining, according to a preset correspondence between a name of a goods batch and information about the two-dimensional code, a name of a goods batch corresponding to the information about the two-dimensional code; determining, according to the name of the goods batch corresponding to the information about the two-dimensional code, information about a plurality of two-dimensional codes corresponding to the name of the goods batch; and associating the information about the plurality of two-dimensional codes with a URL in which goods information is written and that corresponds to the name of the goods batch corresponding to the information about the two-dimensional code.

Abstract: A two-dimensional dot matrix barcode encoding method comprises: forming a plurality of coordinate calibration blocks, a plurality of data storage blocks, a first indication block and a second indication block respectively by filling code points into the virtual code point storage blocks, wherein the first indication block and the plurality of coordinate calibration blocks are filled with a code point located in the center of the virtual code point storage blocks; and forming, by means of a combination of the first indication block and the second indication block, a direction and coordinate indication set indicative of a direction and a coordinate, wherein the second indication block is not filled with a code point or is filled, in the center thereof, with an ink code point capable of reflecting a light having a specific wavelength, and the plurality of coordinate calibration blocks are separated from each other.

Abstract: A two-dimensional dot matrix barcode encoding method comprises: forming a plurality of coordinate calibration blocks, a plurality of data storage blocks and a direction indication block by filling code points into the virtual code point storage blocks, wherein the plurality of coordinate calibration blocks are filled with only one code point located in the center of the virtual code point storage block; the direction indication block is filled with at least two code points, only one code point being located in the center of the virtual code point storage block; the plurality of coordinate calibration blocks and the direction indication block are separated from each other and distributed among the plurality of data storage blocks; and code points filled in the plurality of data storage blocks are used to store data information, where M?4, N?4, N?4, and M and N are both even numbers.

Abstract: A metal smartcard (SC) having a transponder chip module (TCM) with a module antenna (MA), and a card body (CB) comprising two discontinuous metal layers (ML), each layer having a slit (S) overlapping the module antenna, the slits being oriented differently than one another. One metal layer can be a front card body (FCB, CF1), and the other layer may be a rear card body (RCB, CF2) having a magnetic stripe (MS) and a signature panel (SP).

Abstract: Smartcard (SC) having a card body (CB) and a conductive coupling frame antenna (CFA) extending as a closed loop circuit around a periphery of the card body, and also extending inwardly so that two portions of the coupling frame antenna are closely adjacent each other, with a gap therebetween. The gap may extend from a periphery of the card body to a position corresponding with a module antenna (MA) of a transponder chip module (TCM) disposed in the card body, and may function like a slit (S) in a coupling frame (CF). A portion of the coupling frame antenna may be arranged to surround the ISO position of the transponder chip module in the card body. A coupling frame antenna (CFA) may be incorporated onto a module tape (MT) for a transponder chip module (TCM).

Abstract: A user-specific, electroencephalogram data-based secure encryption generator maps artificial neural network neuron elements to electroencephalogram data signals generated from scanning neural activity of a user while the user executes a mental activity. Weighting factors are trained to transform the electroencephalogram data signals into a first set of weighted signals that are different from weighted signals generated from scanning neural activity of the user while the user executes another, different activity, and from weighted signals generated from scanning neural activity of another user while executing a similar mental activity. The trained weighting factors are associated with the first set of electroencephalogram data signals and the current mental activity. Thus, a reproducible electroencephalogram encryption key is defined that is unique to the user as a function of one or both of the trained weighting factors and the first weighted set of electroencephalogram data signals.

Abstract: Memristive devices and methods for setting the resistance of a memristive device include a first mixed conducting layer formed from a first material having a resistance that changes depending on an ion concentration and having multiple coexisting phases from concentration-dependent metastability. A second metastable, mixed conducting layer is formed from the first material. A barrier layer between the first conductor layer and the second conductor layer is formed from a second mixed conducting material having a chemical potential that prevents thermal ion diffusion between the first and second layer. The barrier layer provides an electrical threshold, above which ions are transferred between the first and second layer and below which the resistance of the device is read.

Abstract: A method for predicting preliminary design load cases using an artificial neural network, the method including obtaining, with a processor, historical data corresponding to an existing structural component, the historical data including finite element modeling analysis data and measured test data for the existing structural component, designing the artificial neural network to include a plurality of neurons to form a numerical model for preliminary design load cases of a new structural component, training, with the processor and the historical data for the existing structural component, the artificial neural network to predict the preliminary design load cases for the new structural component, providing input design parameters for the new structural component to the artificial neural network, and predicting, with the artificial neural network, the preliminary design load cases for the new structural component.

Abstract: Optimizing the performance of a machine learning system includes: defining an n-dimensional approximate computing configuration space, the n-dimensional approximate computing configuration space defining tuning parameters for tuning the machine learning system; setting a performance objective for the machine learning system that identifies one or more machine learning system performance criteria; collecting and monitoring performance data; comparing the performance data to the machine learning system performance objective; and dynamically updating the n-dimensional approximate computing configuration space by adjusting the at least one tuning parameter, in response to the comparison.

Abstract: A deep learning system is provided for balancing memory consumption. The deep learning system includes a central processing unit (CPU). The deep learning system further includes a plurality of Graphics Processing Units (GPUs), each having a memory that includes a neural network for neural network training. The memory of a particular one of the plurality of GPUs having a smallest memory consumption includes an additional neural network for neural network validation.

Abstract: An implementation of the disclosure provides an apparatus comprising a memory to store classification data of a machine learning model; and a processing device, operatively coupled to the memory, to detect characteristics of an object. The characteristics comprise at least a spatial position of the object. Sensor information is received from a node device. The sensor information is indicative of at least one of: a spatial position, a speed or a direction of movement of the node device. Using the sensor information, it is determined whether the node device is within a threshold with respect to the object in accordance with the detected characteristics. Responsive to determining that the node device is within the threshold, a notification related to at least part of the classification data and the one or more characteristics is provided to the node device. The classification data is representative of a classification of the object.

Abstract: A method, computer system, and a computer program product for generating a location tag for a piece of visual data using deep learning is provided. The present invention may include receiving the piece of visual data. The present invention may also include analyzing the received piece of visual data using a neural network. The present invention may then include retrieving a location for the analyzed piece of visual data from the neural network. The present invention may further include generating a plurality of metadata for the retrieved location associated with the analyzed piece of visual data, wherein the generated plurality of metadata includes the location tag.