Abstract: A control device in a premises security system for a premises is provided. The control device is configured to receive video surveillance data associated with an area of the premises, identify, using at least one machine learning model, a triggering event based at least in part on the video surveillance data, identify, using the at least one machine learning model, and based at least in part on the video surveillance data, a person associated with the triggering event, identify, using the at least one machine learning model, at least one characteristic of the person associated with the triggering event, generate an audio message comprising content based at least in part on the at least one characteristic of the person associated with the triggering event, and cause playback of the audio message in the area of the premises.

Abstract: A vehicle periphery monitoring device stores images photographed with a front camera, a rear camera, a left side camera, and a right side camera as a past front image, a past rear image, a past left side image, and a past right side image, respectively. When a vehicle travels while making a turn, the vehicle periphery monitoring device generates an underfloor image indicating a condition of an underfloor of the vehicle using at least one of the past front image and the past rear image and at least one of the past left side image and the past right side image. The vehicle periphery monitoring device displays the generated underfloor image on a display.

Abstract: There is provided a camera and a method therein for facilitating installation of the camera. The method comprises capturing a video stream of a scene when the camera is installed with a first FOV, which scene comprises an object. A size in the video stream of at least one part of the object is estimated. As long as the estimated size is above a first threshold and below a second threshold, providing a first, edge-filtered, version of the video stream to secure privacy of the object. In response to the estimated size being above the second threshold indicating that the privacy is at risk of being revealed, entering an enhanced secure privacy mode comprising generating an alert to prompt an installer to re-install the camera with a second FOV different from the first FOV, and disabling provision of the first, edge-filtered, version of the video stream.

Abstract: The disclosure discloses a method of neural network model computation-oriented intermediate representation and apparatus thereof. The method includes the following steps: S1, parsing an input model file so as to acquire topological structure information of a neural network; S2, constructing a logical computation graph; S21, inferring physical layout information of each operator in the logical computation graph; S22, inferring meta attributes of each operator in the logical computation graph; S23, inferring description information of input and output logical tensors of each operator in the logical computation graph; S3, constructing a physical computation graph; S31, generating a physical computation graph, etc.

Abstract: A method includes processing, using at least one processor of an electronic device, multiple reference images of a scene using a first convolutional neural network (CNN) to generate a confidence map and a disparity map. The method also includes generating, using the at least one processor, an initial Bokeh image based on the disparity map and the reference images using a depth-of-field (DoF) renderer. The method further includes refining, using the at least one processor, the initial Bokeh image using a second CNN to generate a refined Bokeh image, where the second CNN uses the confidence map, the disparity map, and the reference images to generate the refined Bokeh image.

Abstract: Disclosed are systems, devices, and methods for detecting a trachea, an exemplary system comprising an imaging device configured to obtain image data and a computing device configured to generate a three-dimensional (3D) model, identify a potential connected component in a first slice image, identify a potential connected component in a second slice image, label the first slice image as a top slice image, label the connected component in the top slice image as an active object, associate each connected component in a current slice image with a corresponding connected component in a previous slice image based on a connectivity criterion, label each connected component in the current slice image associated with a connected component of the preceding slice image as the active object, and identify the active object as the trachea, based on a length of the active object.

Abstract: The present invention relates to a method of detecting an anomaly using a plurality of images. The method comprises determining a subset of images from the plurality of images, comprising one or more first features. Further, one or more regions comprising one or more second features are identified in each image from the subset of images. Finally, detecting the anomaly from the subset of images based on the one or more regions. Therefore, accuracy of detecting anomalies is increased by using the subset of images.

Abstract: The disclosed computer-implemented method may include (1) receiving a present frame of a video stream, the present frame comprising a present depiction of a multi-segment articulated body system, (2) identifying a previous frame of the video stream that comprises a previous depiction of the multi-segment articulated body system, (3) analyzing the present frame and the previous frame to determine whether the multi-segment articulated body system remained substantially rigid between the previous frame and the present frame, and (4) estimating a pose of the multi-segment articulated body system in the present frame using a first pose estimation computation that treats the multi-segment articulated body system as rigid and that is selected in contrast to a second pose estimation computation based on determining that the multi-segment articulated body system remained substantially rigid between the previous frame and the present frame. Various other methods, systems, and computer-readable media are also disclosed.

Abstract: This application belongs to the field of computer technologies, and provides an animation making method and apparatus, a computing device and a storage medium, to improve execution efficiency of animation making. In response to a pose selection instruction for a non-reference skeleton pose, a target plug-in node is invoked, the target plug-in node obtaining a non-reference skeleton shape model corresponding to the non-reference skeleton pose from a non-reference skeleton shape model set based on the pose selection instruction; target skeleton pose for an animated character is determined based on a parameter input instruction for a parameter of the target skeleton pose; and a target skeleton shape model of the target skeleton pose is generated based on the obtained non-reference skeleton shape model of the non-reference skeleton pose.

Abstract: An embodiment for creating a tiled image using different zoom levels based on user interests is provided. The embodiment may include receiving one or more photographs captured by a user. The embodiment may also include analyzing features associated with the one or more photographs. The embodiment may further include identifying one or more known objects of interest. The embodiment may also include in response to determining the user wants to take a tiled zoom photograph, capturing a photographic image. The embodiment may further include scanning the photographic image for identifying features of the one or more known objects of interest. The embodiment may also include in response to determining at least one object of interest is found in the scanned photographic image, assembling the tiled zoom photograph with a plurality of tiles.

Abstract: A system for securely accessing a user account is disclosed. The system includes an environmental data subsystem to receive a set of environmental data representative of an environment from sensors, wherein the set of environmental data includes a first set of environmental data received at a first time instant and a second set of environmental data received at a second time instant; an environment identification subsystem to extract a first set of unique features of the environment from the first set of environmental data and a second set of unique features of the environment from the second set of environmental data; an environmental data comparison subsystem to compare the second set of unique features with the first representation of the first set of unique features, to determine similarities between the second set of unique features and the first representation, to generate a confidence score based on the similarity analysis.

Abstract: Methods and systems are described for registering a sports field to a video. Video of a live event may feature participants at a venue. A template of the venue, including virtual markings that represent real markings on the venue, may be obtained. A homographic transformation between an image plane and a ground plane may be determined by matching virtual markings to corresponding real markings captured in at least one frame of the video. The determined homographic transformation may be used in the automated analysis of sports statistics and in improving inserted annotations and visualizations.

Abstract: A camera system may automatically identify high frequency motion regions in a field of view of the camera system. Analysis of motion events detected at different regions of a scene in the field of view, may be performed. Similar events frequently occurring at a certain region may be recorded/accumulated and compared to a threshold. Regions with high frequency activity may be determined and a user may be notified. If the user is not interested in events within the high activity regions, the regions may be ignored and notifications to the user may cease. The camera system may determine specific actions/events occurring in a region, such as a person approaching or leaving a home.

Abstract: An image restoration apparatus includes: an image acquisition unit that acquires a mask image generated by masking, by a mask region, a predetermined region of an article image in which an article is imaged; and an image output unit that outputs a restored image corresponding to the mask image acquired by the image acquisition unit by using a restorer pre-trained by machine learning so as to output the restored image, which is restored from the mask image to reproduce the article image in a pseudo manner, according to an input of the mask image.

Abstract: In a system for identifying objects by means of distributed neural networks a resource-intensive proportion of the neural network is provided at a base station (on the ground), whereas a less resource-intensive proportion of the neural network, in particular the remaining proportion of the neural network, is provided at a front end (for example remote-controlled vehicle). The technical complexity for the front-end side and for the transmission of data between the base station and front end can thus be reduced.

Abstract: A technique for assessing development condition of a property is provided that determines development condition for an individual property or properties of interest using image or other sensor data from one or more unmanned aerial vehicles taken over the development process. A property condition output may be generated to indicate a condition of the property or properties.

Abstract: The present technology relates to image data processing devices, image data processing methods, and programs. A frame data generation unit generates first frame data based on event data indicating a variation in an electrical signal of a pixel generating the electrical signal by performing photoelectric conversion during a first accumulation time from a first frame generation start time to a first frame generation end time, and second frame data based on event data occurring during a second accumulation time from a second frame generation start time to a second frame generation end time. A first frame period from the first frame generation start time to the second frame generation start time is set and supplied to the frame data generation unit. The present technology can be applied to, for example, a case where a frame data is generated from an event data output from a dynamic vision sensor (DVS).

Abstract: A tracking device comprising at least one memory and at least one processor which function as: an acquisition unit configured to acquire a viewed position, which is a position on a display viewed by a user; a tracking unit configured to track an object displayed on the display; and a control unit configured to perform control processing to control the tracking unit based on a degree of irregularity of a change of the viewed position.

Abstract: A method and system for mapping fluid objects on a substrate using a microscope inspection system that includes a light source, imaging device, stage for moving a substrate disposed on the stage, and a control module. A computer analysis system includes an object identification module that identifies for each of the objects on the substrate, an object position on the substrate including a set of X, Y, and ? coordinates using algorithms, networks, machines and systems including artificial intelligence and image processing algorithms. At least one of the objects is fluid and has shifted from a prior position or deformed from a prior size.

Abstract: The present disclosure provides a method for acquiring an item placing state, including: sending an acquisition request, wherein the acquisition request includes one or more association relationships, each of the association relationships includes an association relationship between a hot zone and a camera equipment, and one hot zone corresponds to one kind of item; and receiving a placing state of an item corresponding to at least one hot zone, wherein the placing state of the item corresponding to the at least one hot zone is determined according to a current image and a standard image of the at least one hot zone in response to the acquisition request. The present disclosure further provides a processing method, a system for acquiring an item placing state and a processing system.

Abstract: A position estimation system includes: a reference image database configured to store a reference image and information indicating a position in a state in which they are associated with each other; an acquisition unit configured to acquire an image for estimation which is a position estimation target; an estimation unit which performs comparison of the reference image with the image for estimation, estimate the position related to the image for estimation from the information indicating the position associated with the reference image, and calculate a reliability of the estimated position; and an update unit configured to update the reference image using the image for estimation on the basis of the reliability.

Abstract: This application relates to a point cloud segmentation method, a computer-readable storage medium, and a computer device. The method includes encoding a to-be-processed point cloud to obtain a shared feature, the shared feature referring to a feature shared at a semantic level and at an instance level; decoding the shared feature to obtain a semantic feature and an instance feature respectively; adapting the semantic feature to an instance feature space and fusing the semantic feature with the instance feature, to obtain a semantic-fused instance feature of the point cloud, the semantic-fused instance feature representing an instance feature fused with the semantic feature; dividing the semantic-fused instance feature of the point cloud, to obtain a semantic-fused instance feature of each point in the point cloud; and determining an instance category to which each point belongs according to the semantic-fused instance feature of each point.

Abstract: Systems and techniques are described herein for modifying the scale and/or position of objects in images. For instance, a system can obtain a two-dimensional (2D) input image from a camera and a three-dimensional (3D) representation of the 2D input image. The system can further determine a first portion of the 3D representation of the 2D input image corresponding to a target object in the 2D input image. The system can adjust a pose of the first portion of the 3D representation of the 2D input image corresponding to the target object. The system can further generate a 2D output image having a modified version of the target object based on the adjusted pose of the first portion of the 3D representation of the 2D input image corresponding to the target object to be output on a display.

Abstract: An autonomous vehicle is disclosed that includes a velocity sensor; a radar system; a digital storage medium; and a controller in communication with the digital storage medium, radar system, and the velocity sensor. The controller, for example, may retrieve map data from the digital storage medium; receive current radar data from the radar system; receive autonomous vehicle velocity data from the velocity sensor; identify radar data points in the current radar data that represent objects in motion; remove radar data points from the current radar data that represent objects in motion; and/or match the radar data with the map data to return location data.

Abstract: A virtual reality encounter system is described. A first surrogate supporting at least one first camera that captures image data from a first physical location and a second surrogate supporting at least one second camera that captures second image data from the first physical location. Aliasing substitution processing has a computing system including a processor receive the first image data and detect an image of the second surrogate in the first image data and replace the image data of the second surrogate in the first physical location, with image data of a user in the first physical location to form a transformed image that substitutes the image data of the user for the image data of the second surrogate.

Abstract: Embodiments of the disclosure disclose a method for a vehicle avoiding an obstacle, an electronic device, and a computer storage medium, and relates to a field of unmanned driving technologies. The detailed implementation includes: obtaining size information of a vehicle; obtaining a safe distance in a traveling direction of the vehicle; determining an obstacle screening area of the vehicle at least based on the size information and the safe distance; and determining an obstacle avoiding path of the vehicle in response to determining that the obstacle is located in the obstacle screening area. With the technical solution of the disclosure, higher efficiency obstacle avoiding can be realized.

Abstract: An action recognition apparatus 1 including: a generation unit 2 that generates an addition image by adding a setting image to a target object image; an action recognition and estimation unit 3 that performs action recognition regarding the target object, outputs action recognition information indicating the result of action recognition, estimates the setting image, and outputs estimation information indicating the result of estimation; an action recognition loss calculation unit 4 that calculates an action recognition loss by using reference action recognition information generated in advance, and the action recognition information; an estimation loss calculation unit 5 that calculates an estimation loss by using reference estimation information generated in advance, and the estimation information; and a learning parameter updating unit 6 that updates learning parameters by using the action recognition loss and the estimation loss.

Abstract: Methods and systems for protecting membership privacy in multi-identification secure computation and communication are provided. The method includes providing a dataset having a first set of membership identifications and a second set of membership identifications, determining a number N based on a data privacy configuration, generating and shuffling a first padding dataset, and up-sampling the first set of membership identifications with a first N elements of the shuffled first padding dataset. The method also includes inserting a first N random membership-identification elements to the second set of membership identifications, generating and shuffling a second padding dataset, up-sampling the inserted second set of membership identifications with a first N elements of the shuffled second padding dataset. The method further includes performing an intersection operation based on the up-sampled dataset and a received dataset.

Abstract: A mechanism is described for facilitating training and deploying of pose regression in neural networks in autonomous machines. A method, as described herein, includes facilitating capturing, by an image capturing device of a computing device, one or more images of one or more objects, where the one or more images include one or more training images associated with a neural network. The method may further include continuously estimating, in real-time, a present orientation of the computing device, where estimating includes continuously detecting a real-time view field as viewed by the image capturing device and based on the one or more images. The method may further include applying pose regression relating to the image capturing device using the real-time view field.

Abstract: Systems and methods for performing visual odometry more rapidly. Pairs of representations from sensor data (such as images from one or more cameras) are selected, and features common to both representations of the pair are identified. Portions of bundle adjustment matrices that correspond to the pair are updated using the common features. These updates are maintained in register memory until all portions of the matrices that correspond to the pair are updated. By selecting only common features of one particular pair of representations, updated matrix values may be kept in registers. Accordingly, matrix updates for each common feature may be collectively saved with a single write of the registers to other memory. In this manner, fewer write operations are performed from register memory to other memory, thus reducing the time required to update bundle adjustment matrices and thus speeding the bundle adjustment process.

Abstract: Techniques are disclosed for training a machine learning model to predict user expression. A plurality of images are received, each of the plurality of images containing at least a portion of a user's face. A plurality of values for a movement metric are calculated based on the plurality of images, each of the plurality of values for the movement metric being indicative of movement of the user's face. A plurality of values for an expression unit are calculated based on the plurality of values for the movement metric, each of the plurality of values for the expression unit corresponding to an extent to which the user's face is producing the expression unit. The machine learning model is trained using the plurality of images and the plurality of values for the expression unit.

Abstract: Multimodal systems are provided for managing safety in an industrial environment. The system comprises: (a) a computer vision component for generating a first output data; (b) a real-time locating component for generating a second output data about an object within the industrial environment and a mobile tag device deployed to the object; (c) a LIDAR component for generating a third output data; and (d) an edge computing device connected to the computer vision component, the real-time locating component and the LIDAR component via a local network, and is configured to: (i) receive a data stream including the first output data, the second output data and the third output data, (ii) process the data stream using a machine learning algorithm trained model to generate a safety related result and feedback data, and (iii) deliver the feedback data to the object via the mobile tag device.

Abstract: The present disclosure provides a pedestrian re-identification method and apparatus. The pedestrian re-identification apparatus detects a pedestrian to be identified in a specified video frame, and within a specified time period before the capture time of the specified video frame, extracts all pedestrians appearing in the captured video frame and taking them to be historical pedestrians; calculating the feature distances between the pedestrian to be identified and the historical pedestrians; in order of feature distance, from small to large, extracting historical pedestrian identifications corresponding to a predetermined number of feature distances, and clustering the extracted historical pedestrian identifications so as to re-identify the pedestrians to be identified. The present disclosure effectively suppresses external noise interference and improves the success rate of pedestrian re-identification.

Abstract: Embodiments of the present systems and methods may provide techniques that may provide the capability to autonomously and automatically identify cases of suspected abuse, which may then be subject to follow up review and action. For example, in an embodiment, a system for identifying potential physical abuse in a location may comprise at least one video camera provided at the location and a computer system comprising a processor, memory accessible by the processor, and computer program instructions stored in the memory and executable by the processor to cause the computer system to perform: receiving video data from the at least one video camera, identifying potential physical abuse from the video data, the identification being to within a predetermined confidence level, and transmitting information indicating that a potential human threat has been identified.

Abstract: A liveness detection method and apparatus, and a facial verification method and apparatus are disclosed. The liveness detection method includes detecting a face region in an input image, measuring characteristic information of the face region, adjusting the measured characteristic information in response to the characteristic information not satisfying a condition, and performing a liveness detection on the face region with the adjusted characteristic information upon the measured characteristic information not satisfying the condition.

Abstract: Generally, the disclosed systems and methods utilize multi-task machine-learned models for object intention determination in autonomous driving applications. For example, a computing system can receive sensor data obtained relative to an autonomous vehicle and map data associated with a surrounding geographic environment of the autonomous vehicle. The sensor data and map data can be provided as input to a machine-learned intent model. The computing system can receive a jointly determined prediction from the machine-learned intent model for multiple outputs including at least one detection output indicative of one or more objects detected within the surrounding environment of the autonomous vehicle, a first corresponding forecasting output descriptive of a trajectory indicative of an expected path of the one or more objects towards a goal location, and/or a second corresponding forecasting output descriptive of a discrete behavior intention determined from a predefined group of possible behavior intentions.

Abstract: The present application provides a hand tracking method, device and system, wherein the method comprises: determining a current frame image corresponding to each tracking camera respectively; acquiring tracking information of a hand location corresponding to the to-be-detected frame image and two-dimensional coordinates of a preset quantity of skeleton points according to the current frame images and the tracking information of the last frame image of the current frame images; determining three-dimensional coordinates of the preset quantity of skeleton points according to the two-dimensional coordinates and pre-acquired tracking data of a head location corresponding to the hand location; carrying out smoothing filter processing on the three-dimensional coordinates of the skeleton points and historical three-dimensional coordinates of the last frame image so as to acquire processed stable skeleton points; and fusing, rendering and displaying the stable skeleton points and the tracking data of the head location

Abstract: A computer-implemented method for generating a rotation-invariant feature descriptor for a location in an image for use in performing descriptor matching in analysing the image, extracts samples according to a descriptor pattern for the location in the image; uses the extracted samples to determine a measure of rotation for the location in the image, the measure of rotation describing an angle between an orientation of the image and a characteristic direction of the image at the location; generating a feature descriptor for the location in the image by determining a set of samples characterising the location in dependence on the determined measure of rotation and the extracted samples; and processes the determined set of samples to generate the feature descriptor for the location in the image.

Abstract: A processor of a server includes: a keyword generator that narrows object images from video based on both activity information on the activity history of a user and a first level of narrowing down and generates candidate keywords, as candidate comments, by generating a candidate keyword for each of the images; an identifier that identifies one or more target keywords from among the candidate keywords generated by the keyword generator based on both the activity information and a second level of narrowing down; a comment generator that generates, for each of the one or more target keywords, a comment related to each target keyword; and an adjuster that adjusts the first level of narrowing down in the keyword generator and the second level of narrowing down in the identifier in accordance with processing information on processing performable by the keyword generator and processing performable by the identifier.

Abstract: Methods, devices, and media are disclosed for mapping of input and output spaces in head-based human-computer interactions. In some embodiments, an end-to-end method is described for designing a head-based user interface, calibrating the interface to individual users, and interacting with a user in real time by mapping head-based user inputs to an output space in a way that optimizes the target selection efficiency of the interaction. Head orientation may be leveraged to define the mapping between the user input and the output space.

Abstract: An apparatus includes a neuromorphic vision unit, an array, and a controller. The neuromorphic vision unit includes a plurality of pixels to photoelectrically produce data based on light received from an object in a scene. The array includes a plurality of light sources. The controller controls the plurality of light sources at one of a plurality of granularities, at least in part based on the data.

Abstract: A skin detection method includes: dividing a region of interest in a face image into a highlighted region and a non-highlighted region; separately determining a first segmentation threshold of the highlighted region and a second segmentation threshold of the non-highlighted region; obtaining a binary image of the highlighted region based on the first segmentation threshold, and obtaining a binary image of the non-highlighted region based on the second segmentation threshold; fusing the binary image of the highlighted region and the binary image of the non-highlighted region; and identifying, based on a fused image, pores and/or blackheads included in the region of interest.

Abstract: A method of monitoring and reporting aircraft turnaround activities at an airport gate that includes monitoring the airport gate using a first camera that has a first field of view associated with the airport gate; generating, using the first camera, first image data capturing a first activity within the first field of view; receiving, by processor(s), the first image data; identifying, by the processor(s), the first activity captured in the first image data; comparing, using a computer application, aspect(s) of the first activity with a predetermined schedule; and displaying, on a graphical user interface remote from the first camera and using the computer application, an alert indicating that the first activity does not comply with the predetermined schedule; wherein the processor(s) include a machine learning processor.

Abstract: A processor-implemented method of tracking a target object includes: extracting a feature from frames of an input image; selecting one a neural network model from among a plurality of neural network models that is provided in advance based on a feature value range, based on a feature value of a target object that is included in the feature of a previous frame among the frames; and generating a bounding box of the target object included in a current frame among the frames, based on the selected neural network model.

Abstract: Disclosed are systems and methods to detect and track an object across frames of a video. One of the disclosed methods includes: detecting a first group of one or more objects, using a first neural network, in each frame of the video, wherein each detected head of the first group comprises a leading and a trailing edge; grouping the leading and trailing edges of the one or more objects into groups of leading edges and groups of trailing edges based at least on coordinates of the leading and trailing edges; generating a list of no-edge-detect frames by identifying frames of the video missing a group of leading edges or a group of trailing edges; analyzing the no-edge-detect frames in the list of no-edge-detect frames, using an optical image classification engine, to detect a second group of one or more objects in the no-edge-detect frames; and merging the first and second groups of one or more objects to form a merged list of detected objects in the video.

Abstract: The disclosed computer-implemented method may include (1) receiving information associated with a real-time output of a recognition model adapted to recognize at least one behavior of a user, (2) determining, based on the information, a level of uncertainty associated with the real-time output, (3) modulating at least one attribute of feedback based on the level of uncertainty, and (4) presenting the feedback to the user substantially contemporaneous with the real-time output of the recognition model. Various other methods, systems, and computer-readable media are also disclosed.

Abstract: One or more embodiments described herein relate to predicting, using adaptive artificial intelligence techniques, typical and aberrant physiological reactions of a patient to psychiatric counseling. Treatment plans can be determined and calculated based on previously-gathered demographic and/or biometric data, and/or modifications to treatment plans can be determined and/or implemented based on emergent recognition of reaction types, such as reclassifying reactions that would previously have been deemed typical as aberrant (or vice versa).

Abstract: A swing analysis system is disclosed herein. The swing analysis system includes a motion capture system comprising at least one motion capture device configured to detect motion of one or more body segments of a user and generate first output data, and further configured to detect the motion of a head and/or face of the user, a hand and/or fingers of the user, and/or an object being manipulated by the user, and generate second output data; and at least one data processing device operatively coupled to the motion capture system, the at least one data processing device configured to determine one or more swing performance parameters for the user using first positional information of the one or more body segments of the user, and/or second positional information for the head and/or face of the user, the hand and/or fingers of the user, and/or the object being manipulated by the user.

Abstract: An incident prediction system and an incident prediction method are provided for predicting an incident. The system comprises a crowd detection interface for receiving a plurality of mobile device identifiers and an incident database wherein a plurality of weight factors is associated with a plurality of stored mobile device identifiers. The weights associated to mobile devices which were present at a site during a previous, historical incident are higher. The system further comprises a prediction subsystem configured for determining a total weight factor at a site to predict an occurrence of an incident. The system and the method may provide information on a composition of a crowd which may help to better prevent incidents from happening. As such, the provided system and method may enable a more effective prediction of incidents compare to currently available systems and methods.

Abstract: Disclosed herein are an apparatus for detecting unknown malware using a variable-length operation code (opcode) and a method using the apparatus. The method includes collecting opcode information from a detection target, generating a multi-pixel image having a variable length by performing feature engineering on the opcode information; and detecting unknown malware by inputting the multi-pixel image to a deep-learning model based on AI.