Methods and apparatus, including computer program products, are provided for watermarking neural networks. In some embodiments, there may be provided a method. The method may include determining, for a neural network, an activation layer output by a hidden layer of the neural network. The method may include selecting a watermarking process. The method may include applying the selected watermarking process to the activation layer output to generate a key. The method may include storing, for the neural network to enable detection of copying of the neural network, the selected watermarking process and the key. Related systems, methods, and articles of manufacture are also described.

This disclosure relates to advanced image signal processing technology including encoded signals and digital watermarking. One implementation is directed to a method comprising: obtaining data representing a color area; determining that the data representing the color area, or information obtained by using the data representing the color area, indicates that the color area comprises a reflectivity of 15 percent (%) or less at or around 660 nm; generating an encoded signal, the encoded signal comprising a pattern including signal elements at various spatial locations corresponding to the color area; generating holes in the color area according to the pattern, said generating holes yield an encoded color area; and selecting a fill color for placement within the holes of the encoded color area, said selecting utilizing an encoded signal error that is associated with the color area, the fill color, and a reflectance difference at or around 660 nm between the color area and the fill color, said selecting also utilizing a color error that is associated with the color area, the fill color, and the reflectance difference. Of course, other implementations, methods, packages, systems and apparatus are described in this patent document.

In one embodiment, a method for 3D digital watermarking for a triangular mesh using one or more key parameters is disclosed including forming a Hamiltonian path of a desired length around a selected vertex in a selected direction of a spiral; marking the selected vertex a dead end if there is a deadlock and continuing the spiral; and applying a watermark by introducing points in a path order on edges of the spiral, wherein information is encoded at a partition of adjacent triangles at one or more of the points.

This disclosure relates to advanced signal processing technology including signal encoding and digital watermarking. Image areas are selected in an encoded digital design, and corresponding areas from a printed version of the encoded digital design are evaluated to determined signal robustness after printing. One claim recites an image processing method for selecting image areas to test for robustness of encoded signals, the method comprising: obtaining digital artwork comprising a plurality of colors, the digital artwork comprising multiple instances of an encoded signal; selecting a set encoding tiles, and for each encoding tile determining encoding detectability measure associated therewith; creating a bin for each encoding technology used to encode the encoded signal; for each bin, removing any encoding tile having a detectability measure below a predetermined threshold; for each bin, prioritizing remaining encoding tiles; selecting an encoding tile based on the prioritization per bin, and spatially locating the selected encoding tile relative to the digital artwork. Other technology is described in this patent document.

An image forming apparatus includes an instruction-information acquiring unit, a translation-result-information acquiring unit, a font-data acquiring unit, an image-formation control unit, and an image forming unit. The instruction-information acquiring unit acquires information indicating a first character string described in a first language and information indicating a second language different from the first language. The translation-result-information acquiring unit acquires, from a translation service provision server, information indicating a second character string, which is a character string obtained by translating the first character string into the second language. The font-data acquiring unit acquires font data of the second language from a font data distribution server. The image-formation control unit generates a synthetic image obtained by adding the second character string having a font based on the font data to a formation target image.

A computer implemented method for embedding a marker in an image or video content including receiving an input image or frame for embedding, determining a binary message to be encoded within said input image or frame comprising bits sequences having an identical number of bits which is superior or equal to two, said binary message comprising at least a header part comprising at least two consecutive bits sequences which are not identical, detecting a region within said input image or frame such that the color within said region is uniform and that said region presents a chosen length and height, associating each possible bits sequence to a corresponding encoding color determined from the color within said uniform region and an encoding rule such that the respective colors are all different from one another, and generating a marker color table in which each element stores an encoding color associated to a bits sequence of the binary message, such that the color table constitutes a color encoding of the binary message, and embedding said marker in said region by appending directionally pixel blocks comprising at least a chosen number of pixels in an appending direction, the pixels within a given pixel block being each colored with the encoding color of an element of the marker color table, each element of the marker color table being associated with at least one pixel block.

Optical code signal components are generated and then transformed into signal bearing art that conveys machine readable data. The components of an optical code are optimized to achieve improved signal robustness, reliability, capacity and/or visual quality. An optimization program can determine spatial density, dot distance, dot size and signal component priority to optimize robustness. An optical code generator transforms tiles of an optical code or image embedded with the optical code into signal-bearing art using stipple, Voronoi, Delaunay or other graphic drawing methods so as to retain prioritized components of the optical code. The optical code is merged into a host image, such as imagery, text and graphics of a package or label, or it may be printed by itself, e.g., on an otherwise blank label or carton. A great number of other features and arrangements are also detailed.

This disclosure relates to advanced signal processing technology including signal encoding and image processing. One implementation describes an encoding system including a masking module. The masking module scales or eliminates signal encoding adjustments based on an image's luminance or chrominance values. Of course, other implementations, combinations and claims are also provided.

The present disclosure relates generally to signal encoding for elements within PDF files. One implementation encodes an artwork element under different encoding conditions, and selects a winner version based on resulting signal robustness and/or visibility. Other implementations generate PDF layer masks to help determine overall embedding robustness, including interference from layered elements. Other implementations are provided too.

This disclosure relates to advanced image signal processing technology including encoded signals and digital watermarking. One implementation is directed to a printed object comprising: a white substrate or background comprising a first area; an ink mixture printed at a first plurality of spatial locations within the first area, the ink mixture printed such that the first area comprises a second plurality of spatial locations without the ink mixture, the ink mixture comprising extender white and Green 7 ink, the ink mixture comprising a volume or weight ratio of 97.5% to 99.75% white extender and 2.5%-0.25% Green 7 ink; in which the first plurality of spatial locations is arranged in a pattern conveying an encoded signal, and in which the white substrate or background and the ink mixture comprise a spectral reflectivity difference at or around 660 nm in a difference range of 8%-30%. Of course, other implementations, methods, packages, systems and apparatus are described in this patent document.