Abstract: Techniques for anomaly detection are described. An exemplary method includes receiving a request to create a training data set from at least one image, the request to include an indication of the at least one image and at least one indication of an operation to perform on the at least one image to generate a plurality of images from the at least one image; creating a training dataset by extracting one or more chunks from a first at least one image according to the request; and receiving one or more requests to train an anomaly detection machine learning model using the created training dataset; and training an anomaly detection machine learning model according to one or more requests using the created training data.

Abstract: Techniques for creating machine learning models for direct homography regression for image rectification are described. In certain embodiments, a training service trains an algorithm on a source view of a training image and a homography matrix of the training image into a machine learning model that generates a normalized homography matrix for an input of the source view. The normalized homography matrix may then be utilized to generate a target view of an image input into the machine learning model. The target view of the image may be used in a document processing pipeline for document images captured using cameras.

Abstract: Techniques for anomaly detection are described. An exemplary method includes receiving one or more requests to train an anomaly detection machine learning model using feedback-based training, the request to indicate one or more of a type of analysis to perform, a model selection indication, and a configuration for a training dataset; training the anomaly detection machine learning model according to the one or more requests using the training data; performing feedback-based training on the trained anomaly detection machine learning model; and using the retrained anomaly detection machine learning model.

Abstract: Techniques for anomaly detection are described. An exemplary method includes receiving a request to create a training data set from at least one image, the request to include an indication of the at least one image and at least one indication of an operation to perform on the at least one image to generate a plurality of images from the at least one image; creating a training dataset by extracting one or more chunks from a first at least one image according to the request; and receiving one or more requests to train an anomaly detection machine learning model using the created training dataset; and training an anomaly detection machine learning model according to one or more requests using the created training data.

Abstract: Techniques for feedback-based training are described.

Abstract: A method and an apparatus for encoding a picture are disclosed. For at least one block of a picture to encode, a block predictor is determined (22) for a decoded first component (21) of said at least one block, from a reconstructed region of a first component of said picture. At least one second component of said at least one block is then encoded (23) by predicting said at least one second component from a second component of said block predictor. Corresponding decoding method and apparatus are disclosed.

Abstract: A content stream comprising video and synchronized illumination data is based on a reference lighting setup from, for example, the site of the content creation. The content stream is received at a user location where the illumination data controls user lighting that is synchronized with the video data, so that when the video data is displayed the user's lighting is in synchronization with the video. In one embodiment, the illumination data is also synchronized with events of a game, so that a user playing games in a gaming environment will have his lighting synchronized with video and events of the game. In another embodiment, the content stream is embedded on a disk.

Abstract: The present principles are directed to a parameterized OETF/EOTF for processing images and video. The present principles provide a method for encoding a picture, comprising: applying a parameterized transfer function to a luminance (L) signal of the picture to determine a resulting V(L) transformed signal; encoding the resulting V(L); wherein the parameterized transfer function is adjusted based on a plurality of parameters to model one of a plurality of transfer functions. The present principles also provide for a method for decoding a digital picture, the method comprising: receiving the digital picture; applying a parameterized transfer function to the digital picture to determine a luminance (L) signal of the digital picture, the parameterized transfer function being based on a plurality of parameters; wherein the parameterized transfer function is adjusted based on a plurality of parameters to model one of a plurality of transfer functions.

Abstract: A content stream comprising video and synchronized illumination data is based on a reference lighting setup from, for example, the site of the content creation. The content stream is received at a user location where the illumination data controls user lighting that is synchronized with the video data, so that when the video data is displayed the user's lighting is in synchronization with the video. In one embodiment, the illumination data is also synchronized with events of a game, so that a user playing games in a gaming environment will have his lighting synchronized with video and events of the game. In another embodiment, the content stream is embedded on a disk.

Abstract: A spatial guided prediction technique uses reconstructed pixels of a first component of a digital video image block to determine prediction modes used to recursively build prediction blocks for the other components of the same digital video image block. The technique builds improved predictions resulting in smaller prediction residuals and less bits to code for a given image quality. In one embodiment, the prediction blocks for the subsequent digital video component blocks are built recursively line by line. In another embodiment, the prediction blocks for subsequent digital video component blocks are built recursively column by column.

Abstract: A method and an apparatus for encoding a picture are disclosed. For at least one block of a picture to encode, a block predictor is determined (22) for a decoded first component (21) of said at least one block, from a reconstructed region of a first component of said picture. At least one second component of said at least one block is then encoded (23) by predicting said at least one second component from a second component of said block predictor. Corresponding decoding method and apparatus are disclosed.

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: The present principles are directed to a parameterized OETF/EOTF for processing images and video. The present principles provide a method for encoding a picture, comprising: applying a parameterized transfer function to a luminance (L) signal of the picture to determine a resulting V(L) transformed signal; encoding the resulting V(L); wherein the parameterized transfer function is adjusted based on a plurality of parameters to model one of a plurality of transfer functions. The present principles also provide for a method for decoding a digital picture, the method comprising: receiving the digital picture; applying a parameterized transfer function to the digital picture to determine a luminance (L) signal of the digital picture, the parameterized transfer function being based on a plurality of parameters; wherein the parameterized transfer function is adjusted based on a plurality of parameters to model one of a plurality of transfer functions.

Abstract: The present disclosure relates to image recognition or image searching. More precisely, the present disclosure relates to pruning local descriptors extracted from an input image. The present disclosure proposes a system, method and device directed to the pruning of local descriptors extracted from image patches of an input image. The present disclosure prunes local descriptors assigned to a codebook cell, based on a relationship of the local descriptor and the assigned codebook cell. The present disclosure includes assigning a weight value for use in pruning based on the relationship of the local descriptor and the assigned codebook cell. This weight value is then used during the encoding of the local descriptors for use in image searching or image recognition.

Abstract: A method for retrieving at least one search image matching a query image commences by first extracting a set of search images. The query image is encoded into a query image feature vector and the search images are encoded into search image feature vectors using an optimized encoding process that makes use of learned encoding parameters. The Euclidean distances between the query image feature vector and the search image feature vectors are then computed. The search images are ranked based on the computed distances; and at least one highest-ranked search image is retrieved.

Abstract: A technique for improving the performance of image classification systems is proposed which consists of learning an adaptation architecture on top of the input features jointly with linear classifiers, e.g., SVM. This adaptation method is agnostic to the type of input feature and applies either to features built using aggregators, e.g., BoW, FV, or to features obtained from the activations or outputs from DCNN layers. The adaptation architecture may be single (shallow) or multi-layered (deep). This technique achieves a higher performance compared to current state of the art classification systems.

Abstract: A method for processing in an encoder, the method comprising receiving, by the encoder, a set of local descriptors derived from an image, obtaining, by the encoder, K code words, wherein K>1; and determining, by the encoder, a first element of a bag-of-words image feature vector by using a differentiable function having a difference between each of the local descriptors and one of the K code words as a first parameter, wherein each of the K code words is used in the differentiable function for determining a different element of the bag-of-words image feature vector.

Abstract: A temporal section that is defined by boundary images is selected in a video sequence. A maximum of k stable image frames are selected in the temporal section of image frames having a lowest temporal activity. Image fingerprints are computed from the selected stable image frames. A mega-frame image fingerprint data structure is constructed from the computed fingerprints.

Abstract: An annotation display assistance device includes a display processing unit that displays images that are possibly included in a first group in relation to a search word; an input accepting unit that accepts a selection of an image that should be included in a second group in relation to the search word among the images displayed by the display processing unit; and an annotation adding unit that detects an image that possibly belongs to the second group among the images displayed by the display processing unit based on the selected image to be included in the second group for adding an annotation associated with the image; wherein the display processing unit displays the annotation associated with the image that is detected to possibly belong to the second group by the annotation adding unit, the annotation indicating that the image possibly belongs to the second group.