Abstract: Motivated by the ability of humans to quickly and accurately detect visual artifacts in images, a neural network model is trained to identify and locate visual artifacts in a sequence of rendered images without comparing the sequence of rendered images against a ground truth reference. Examples of visual artifacts include aliasing, blurriness, mosaicking, and overexposure. The neural network model provides a useful fully-automated tool for evaluating the quality of images produced by rendering systems. The neural network model may be trained to evaluate the quality of images for video processing, encoding, and/or compression techniques. In an embodiment, the sequence includes at least four images corresponding to a video or animation.

Abstract: Systems and methods for image processing are described. One or more embodiments of the present disclosure identify a latent vector representing an image of a face, identify a target attribute vector representing a target attribute for the image, generate a modified latent vector using a mapping network that converts the latent vector and the target attribute vector into a hidden representation having fewer dimensions than the latent vector, wherein the modified latent vector is generated based on the hidden representation, and generate a modified image based on the modified latent vector, wherein the modified image represents the face with the target attribute.

Abstract: Proposed is a technique for performing sexing of a fertilized egg with a high sexing rate at a high speed, based on image data obtained by photographing a contour of the egg. In particular, a fertilized egg sexing method, a fertilized egg sexing apparatus and a program for making a female/male judgment of a fertilized egg according to a contour of the fertilized egg by a computer are proposed. The fertilized egg sexing method includes: extracting the contour based on image data obtained by photographing the fertilized egg at different angles, calculating a short radial from the contour, calculating short radial phase differences corresponding to photographing at the angles, and making a female/male judgment using the short radial phase differences.

Abstract: A solid-state imaging device includes: an imager (11) configured to acquire image data; a processing unit (14) configured to execute, on the image data or data based on the image data, processing of extracting a specific region based on a neural network calculation model; and an output unit (16) configured to output image data fabricated based on the specific region or the image data read from the imager based on the specific region.

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: The satellite image processing method includes: acquiring a first target satellite image; defogging the first target satellite image through a first neural network to acquire a first satellite image; and adjusting an image quality parameter of the first satellite image through a second neural network to acquire a second satellite image.

Abstract: A method for modularizing high dimensional neural networks into neural networks of lower input dimensions. The method is suited to generating full-DOF robot grasping actions based on images of parts to be picked. In one example, a first network encodes grasp positional dimensions and a second network encodes rotational dimensions. The first network is trained to predict a position at which a grasp quality is maximized for any value of the grasp rotations. The second network is trained to identify the maximum grasp quality while searching only at the position from the first network. Thus, the two networks collectively identify an optimal grasp, while each network's searching space is reduced. Many grasp positions and rotations can be evaluated in a search quantity of the sum of the evaluated positions and rotations, rather than the product. Dimensions may be separated in any suitable fashion, including three neural networks in some applications.

Abstract: Two or more aerial vehicles, such as drones, can each include one or more cameras for capturing surveying or other image data within a field of view. Some of the image data, as well as other data, can be stored and pre-processed on the vehicles, including via onboard edge computing devices. The image and/or other data can be analyzed by a remote system or service, which has additional computing resources, and used to determine the occurrence or existence of an anomaly, such as a seismic event or structure identified in the field of view. The anomaly determination may be based on a disparity, involving a distance between cameras, the cameras' focal length, and a perpendicular distance from a point between cameras to a surface or subsurface point.

Abstract: Techniques are disclosed for using and training a descriptor network. An image may be received and provided to the descriptor network. The descriptor network may generate an image descriptor based on the image. The image descriptor may include a set of elements distributed between a major vector comprising a first subset of the set of elements and a minor vector comprising a second subset of the set of elements. The second subset of the set of elements may include more elements than the first subset of the set of elements. A hierarchical normalization may be imposed onto the image descriptor by normalizing the major vector to a major normalization amount and normalizing the minor vector to a minor normalization amount. The minor normalization amount may be less than the major normalization amount.

Abstract: Methods, apparatus, systems and articles of manufacture to defend against adversarial machine learning are disclosed. An example apparatus includes memory; computer readable instructions; and processor circuitry to execute the computer readable instructions to: generate a first output indicating a feature that contributed to the generation of a classification by a machine learning model; compare the first output with a second output generated by a server that trained the machine learning model; and flag the machine learning model as corresponding to at least one of model drift or an adversarial attack when first output differs from the second output by more than a threshold.

Abstract: An electronic device includes a device housing having a front side and a rear side, a first image capture device positioned on the front side, and a second image capture device positioned on the rear side. One or more processors of the electronic device cause, in response to user input received at a user interface requesting the second image capture device capture an image of an object, the first image capture device to capture another image of a user delivering the user input. The one or more processors then define an image orientation of the image of the object to be the same as another image orientation of the other image of the user.

Abstract: A computer-implemented method executed by at least one processor for detecting tattoos on a human body is presented. The method includes inputting a plurality of images into a tattoo detector, selecting one or more images of the plurality of images including tattoos, extracting, via a feature extractor, tattoo feature vectors from the tattoos found in the one or more images of the plurality of images including tattoos, applying a deep learning tattoo matching model to determine potential matches between the tattoo feature vectors and preexisting tattoo images stored in a tattoo training database, and generating a similarity score between the tattoo feature vectors and one or more of the preexisting tattoo images stored in the tattoo training database.

Abstract: Video camera or video camera module, comprising an RGB image sensor, a computer processor adapted to perform pre-processing followed by video compression, and a dazzle detector, characterised in responding to dazzle in one colour channel by reducing values in that colour channel prior to compressing the video data. This has the advantage that in the event of laser dazzle that is specific to a colour channel, the compressed video data generated will retain more detail from the other colour channels compared to a conventional camera.

Abstract: Methods and systems for performing concept-based adversarial generation with steerable and diverse semantics. One system includes an electronic processor configured to access an input image. The electronic processor is also configured to perform concept-based semantic image generation based on the input image. The electronic processor is also configured to perform concept-based semantic adversarial learning using a set of semantic latent spaces generated as part of performing the concept-based semantic image generation. The electronic processor is also configured to generate an adversarial image based on the concept-based semantic adversarial learning. The electronic processor is also configured to test a target model using the adversarial image.

Abstract: This application provides an article recommendation method and apparatus, a computer device, and a storage medium.

Abstract: Convolutional neural networks for detecting objects of interest within images of biological specimens are disclosed. Also disclosed are systems and methods of training and using such networks, one method including: obtaining a sample image and at least one of a set of positive points and a set of negative points, wherein each positive point identifies a location of one object of interest within the sample image, and each negative point identifies a location of one object of no-interest within the sample image; obtaining one or more predefined characteristics of objects of interest and/or objects of no-interest, and based on the predefined characteristics, generating a boundary map comprising a positive area around each positive point the set of positive points, and/or a negative area around each negative point in the set of negative points; and training the convolutional neural network using the sample image and the boundary map.

Abstract: An automated system for mitigating risk from a wind farm. The automated system may include an array of a plurality of image capturing devices independently mounted in a wind farm. The array may include a plurality of low resolution cameras and at least one high resolution camera. The plurality of low resolution cameras may be interconnected and may detect a spherical field surrounding the wind farm. A server is in communication with the array of image capturing devices. The server may automatically analyze images to classify an airborne object captured by the array of image capturing devices in response to receiving the images.

Abstract: Systems and methods are disclosed for capturing multiple sequences of views of a three-dimensional object using a plurality of virtual cameras. The systems and methods generate aligned sequences from the multiple sequences based on an arrangement of the plurality of virtual cameras in relation to the three-dimensional object. Using a convolutional network, the systems and methods classify the three-dimensional object based on the aligned sequences and identify the three-dimensional object using the classification.

Abstract: Apparatus and methods for the stitch zone calculation of a generated projection of a spherical image. In one embodiment, a computing device is disclosed which includes logic configured to: obtain a plurality of images; map the plurality of images onto a spherical image; re-orient the spherical image in accordance with a desired stitch line and a desired projection for the desired stitch line; and map the spherical image to the desired projection having the desired stitch line. In a variant, the desired stitch line is mapped onto an optimal stitch zone, the optimal stitch zone characterized as a set of points that defines a single line on the desired projection in which the set of points along the desired projection lie closest to the spherical image in a mean square sense.

Abstract: A system and method is disclosed for plant maintenance, identification and viability scoring. An application is located on consumer devices connected to the server. The application, operating on a smart phone, utilizes onboard GPS, user input, and camera subsystems to customize plant care tips specific to a yard location and plant type. Images may be submitted to the server to identify a plant type through convolutional neural network image recognition. The invention uses another artificial neural network to predict a plant's viability score. The server receives input, such as plant type, soil type, yard location, and amount of sunlight, and the server retrieves local climactic data and plant type optimal values to return the plant's viability score for the selected location. Another aspect of the invention generates and displays an augmented reality display of a plant in the user's yard.