Abstract: System, device, and method for improved encoding and enhanced compression of images, videos, and media content. A method includes: (a) receiving a source image, and analyzing its content on a pixel-by-pixel basis, and classifying each pixel as either (I) a pixel associated with Photographic content, or (II) a pixel associated with Non-Photographic content; (c) generating a pixel-clusters map that indicates (i) clusters of pixels that were classified as Photographic content, and (ii) clusters of pixels that were classified as Non-Photographic content; (d) generating a composed image, by: (d1) applying a first encoding technique, particularly lossy encoding, to encode pixel-clusters that were classified as Photographic content; (d2) applying a second, different, encoding technique, particularly lossless encoding, to encode pixel-clusters that were classified as Non-Photographic content.

Abstract: Method are provided that exhibit increased quality and compression factor for compressing images. The methods can include generating a set of coefficients indicative of image contents of a block of image pixels at a plurality of spatial frequencies. The set of coefficients is scaled to generate a first set of scaled coefficients. An assessment is performed for a plurality of quantization levels, which includes quantizing a subset of the first set of scaled coefficients according to respective quantization levels to generate a quantized subset of the first set of scaled coefficients and determining a post-quantization energy of the quantized subset of the first set of scaled coefficients. Based on the assessment of the plurality of quantization levels, a scaled and quantized version of the set of coefficients is generated. An encoded version of the image based on the scaled and quantized version of the set of coefficients is generated.

Abstract: One example method includes gathering, by a domain adversarial neural network model deployed in an autonomous vehicle operating in a domain, a dataset that comprises unsegmented and unlabeled image data about the domain, sampling the dataset to create an adapted domain dataset, detaching a domain classifier from the domain adversarial neural network, using the domain classifier as a domain change detector model to predict a class of the unsegmented and unlabeled image data in the adapted domain dataset, and based on the class, either: determining that the domain is changed or is unknown; or, determining that the domain has not changed.

Abstract: An image capturing method has: providing an image capturing area on a display screen of a user device; providing an indication area in the image capturing area; marking a license plate after identifying the license plate based on at least one license plate feature in the image capturing area; determining whether the marked license plate is located in the indication area and presented in a predetermined ratio; and capturing an image including the license plate in the image capturing area after the marked license plate is located in the indication area and presented in a predetermined ratio.

Abstract: One embodiment provides a method, the method including: receiving, at a shock analysis system, image data corresponding to at least one image capture device and associated with transportation of a package; determining, at the shock analysis system and by analyzing the image data, the package has undergone a shock of a magnitude exceeding a predetermined threshold; and performing, at the shock analysis system and responsive to determining the package has undergone the shock exceeding the predetermined threshold, at least one predetermined action, wherein the at least one predetermined action comprises notifying a user of the shock.

Abstract: An information processing apparatus includes: an acquisition unit that acquires height information relating to a height of a capturing position of an imaging device that captured an image in which a face of a user was detected; and a determination unit that determines, based on the height information, whether or not the user is a candidate for person requiring assistance, who may be a person requiring assistance.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an encoder neural network configured to receive a data item and to process the data item to output a compressed representation of the data item. In one aspect, a method includes, for each training data item: processing the data item using the encoder neural network to generate a latent representation of the training data item; processing the latent representation using a hyper-encoder neural network to determine a conditional entropy model; generating a compressed representation of the training data item; processing the compressed representation using a decoder neural network to generate a reconstruction of the training data item; processing the reconstruction of the training data item using a discriminator neural network to generate a discriminator network output; evaluating a first loss function; and determining an update to the current values of the encoder network parameters.

Abstract: A method for increasing the confidence of a match between a test image and an image stored in a library database in which features are extracted from the test image and compared to features stored in the image database. If a match is determined, one or more transformations are performed on the test image to generate pose-altered images from which features are extracted and matched with pose-altered images in the database. The scores for the subsequent matchings can be aggregated to determine an overall probability of a match between the test image in an image in the library database.

Abstract: A system (700) including a media comparison engine (MCE) (507) and a method for determining matches between frames in media assets for automating downstream media workflows are provided. For a target master frame in each boundary of master shots of a master media asset (MMA), the MCE (507) performs a comparison with each source frame from each source media asset (SMA), in an optimal search space, using computed signatures; determines matches in the optimal search space; computes a rate of information exchange (RIE) score for each match based on the signature(s) and similarity scores of each source frame in a final search space; identifies a best match based on the RIE score of each match; performs a comparison of source frames subsequent to the best matching source frame with master frames subsequent to the target master frame.

Abstract: Apparatuses, systems, and techniques are presented to generate ultrasound images. In at least one embodiment, use one or more neural networks are used to generate one or more ultrasound images of one or more objects based, at least in part, upon one or more acoustic properties of the one or more objects.

Abstract: Systems, methods, tangible non-transitory computer-readable media, and devices for detecting objects are provided. For example, the disclosed technology can obtain a representation of sensor data associated with an environment surrounding a vehicle. Further, the sensor data can include sensor data points. A point classification and point property estimation can be determined for each of the sensor data points and a portion of the sensor data points can be clustered into an object instance based on the point classification and point property estimation for each of the sensor data points. A collection of point classifications and point property estimations can be determined for the portion of the sensor data points clustered into the object instance. Furthermore, object instance property estimations for the object instance can be determined based on the collection of point classifications and point property estimations for the portion of the sensor data points clustered into the object instance.

Abstract: For updating a computer vision model, a method converts a user input drawing including a user annotation of a first image in a drawing format to a training format image in a training format for a computer vision model. The method generates a training-representation drawing from the training format image. The training-representation drawing includes an image inference for the first image. The method receives user feedback for the training-representation drawing in the drawing format. The method updates the computer vision model based on the user feedback. The method generates an image inference for a second image based on the updated computer vision model and generates model-health metrics, agreement-metrics, and a sortable image index to explain image inferences with respect to guided user annotation of the second image. The method caches partial results from the image-inferences to afford quicker updating of computer vision models, affording more iterative model-development than ad-hoc model-evaluation.

Abstract: Bit-flip object insertion techniques are provided for use with a non-volatile memory (NVM) wherein an object is inserted into a background image by flipping or inverting one or more bits within the pixels of the background image that correspond to the shape and insertion location of an object being inserted. In an illustrative example, pixels within the background image that correspond to the shape and insertion location of the object are XORed with binary 1s. This flips the bits of those pixels to change the color (hue) and/or intensity (brightness) of the pixels so the object appears in the background image. In other examples, only the most significant bits of pixels in the background image are inverted (flipped). Exemplary latch-based procedures are described herein for high-speed processing on an NVM die. Multiple plane NVM die implementations are also described for massive processing.

Abstract: A system comprises an encoder configured to compress and encode data for a three-dimensional mesh using a video encoding technique. To compress the three-dimensional mesh, the encoder determines sub-meshes and for each sub-mesh: texture patches and geometry patches. Also the encoder determines patch connectivity information and patch texture coordinates for the texture patches and geometry patches. The texture patches and geometry patches are packed into video image frames and encoded using a video codec. Additionally, the encoder determines boundary stitching information for the sub-meshes. A decoder receives a bit stream as generated by the encoder and reconstructs the three-dimensional mesh.

Abstract: A dependency indication is signaled within the beginning of a packet, that is, within the adjacent of a slice header to be parsed or a parameter set. This is achieved, for example, by including the dependency indication at the beginning of the slice header, preferably after a syntax element identifying the parameter set and before the slice address, by including the dependency indication before the slice address, by providing the dependency indication to a NALU header using a separate message, or by using a special NALU type for NALUs carrying dependent slices.

Abstract: An application processor includes a neural processing unit configured to convert an input image into a first image based on a first pattern and generate a second image using a neural network, the second image compensating for the conversion; and an image signal processor including a plurality of pipelines configured to perform image signal processing, the plurality of pipelines including at least a first pipeline configured to receive the first image and a second pipeline configured to receive the second image.

Abstract: Various examples relate to determining a number and/or a confluency of cells in a microscopy image. To that end, the microscopy image is firstly rescaled and then processed.

Abstract: In an object detection device, a plurality of object detection units output a score indicating probability that a predetermined object exists, for each partial region set to image data inputted. The weight computation unit computes weights for merging the scores outputted by the plurality of object detection units, using weight calculation parameters, based on the image data. The merging unit merges the scores outputted by the plurality of object detection units, for each partial region, with the weights computed by the weight computation unit. The target model object detection unit configured to output a score indicating probability that the predetermined object exists, for each partial region set to the image data. The first loss computation unit computes a first loss indicating a difference of the score of the target model object detection unit from a ground truth label of the image data and the score merged by the merging unit.

Abstract: The present disclosure relates to a method comprising: providing a trained machine learning model. The trained machine learning model is configured for reconstructing images from input data. The method comprises: receiving (201) a multidimensional array comprising M dimensional acquired data; determining (205) a subset of values of at least one K selected dimension of the array; for each value of the subset determining (207) a M?K dimensional array comprising the acquired data corresponding to the value, resulting in a set of M?1 dimensional arrays; inputting (209) the set of M?K dimensional arrays to the trained machine learning model, and receive a reconstructed image from the trained machine learning model.

Abstract: A method comprising: receiving, as input, training images, wherein at least a majority of the training images represent normal data instances; receiving, as input, a target image; extracting (i) a set of feature representations from a plurality of image locations within each of the training images, and (ii) target feature representations from a plurality of target image locations within the target image; calculating, with respect to a target image location of the plurality of target image locations in the target image, a distance between (iii) the target feature representation of the target image location, and (iv) a subset from the set of feature representations comprising the k nearest the feature representations to the target feature representation; and determining that the target image location is anomalous, when the calculated distance exceeds a predetermined threshold.