Abstract: Techniques are disclosed for implementing a neural network that outputs embeddings. Furthermore, techniques are disclosed for using sensor data to train a neural network to learn such embeddings. In some examples, the neural network may be trained to learn embeddings for instance segmentation of an object based on an embedding for a bounding box associated with the object being trained to match pixel embeddings for pixels associated with the object. The embeddings may be used for object identification, object matching, object classification, and/or object tracking in various examples.

Abstract: In generation of training data for image recognition, in order to reduce a user's workload of assigning a label to a captured image, a learning device includes a processor for performing operations to generate training data and a display device, wherein the processor is configured to acquire a captured image from an image capturing device; acquire one or more candidate objects recognized based on an identification model for a recognition target included in the captured image; and display, on the display device, information on the candidate objects as respective label candidates for the captured image.

Abstract: A histogram-based method for auto segmentation of integrated circuit structures is disclosed. The method includes an auto-segmentation process/algorithm, which works on the histogram of the SEM image and does not try to model the noise sources or the features. The auto-segmentation process/algorithm extracts the number of peaks in the histogram from low magnification SEM images or SEM images not necessarily having high quality images, significantly simplifies the traditionally lengthy and expensive IC reverse engineering efforts. Hence, the size of the image does not affect the final segmentation. The auto-segmentation process/algorithm performs the steps of: extract a first histogram from the first SEM image; identifying boundaries of the plurality of structural elements in the IC based at least in part on an output of the first histogram; and auto-segmenting the first SEM image into the plurality of structural elements.

Abstract: Provided is a non-invasive system and method of determining muscle tissue quality based on image processing. The non-invasive system and method includes determining muscle intramuscular fat content. The methods includes receiving at least one ultrasound scan image of at least a portion of a skin layer as disposed above one or more additional tissue layers, the skin layer defining a horizontal axis and the image provided by a plurality of pixels. The method continues by blurring the pixels of the image and thresholding the pixels of the image to provide an image having a plurality of structural elements of different sizes and gray scale. The method continues with morphing the structural elements of the image to remove small structural elements and connect large structural elements. With this resulting image, the method distinguishes muscle tissue from remaining elements. A ratio of black to white elements is evaluated to determine the muscle tissue quality or intramuscular fat content.

Abstract: In non-limiting examples of the present disclosure, systems, methods and devices for optimizing machine-learned tree ensemble prediction systems are presented. A plurality of instances may be processed by a tree ensemble. Determinations regarding the expected output values of one or more nodes of the tree ensemble may be made based, at least in part, on the processed instances. Further determinations regarding the node contribution values of one or more nodes of the tree ensemble to downstream nodes may be made based on the node output values. In some examples, feature value ranges may be computed for one or more features of the tree ensemble. One or more tree ensemble optimization operations may be performed based on the information determined from one or more of the above-described operations.

Abstract: An acceleration and compression method for a deep convolutional neural network based on quantization of a parameter provided by the present application comprises: quantizing the parameter of the deep convolutional neural network to obtain a plurality of subcode books and respective corresponding index values of the plurality of subcode books; acquiring an output feature map of the deep convolutional neural network according to the plurality of subcode books and respective corresponding index values of the plurality of subcode books. The present application may implement the acceleration and compression for a deep convolutional neural network.

Abstract: Techniques are disclosed for implementing a neural network that outputs embeddings. Furthermore, techniques are disclosed for using sensor data to train a neural network to learn such embeddings. In some examples, the neural network may be trained to learn embeddings for instance segmentation of an object based on an embedding for a bounding box associated with the object being trained to match pixel embeddings for pixels associated with the object. The embeddings may be used for object identification, object matching, object classification, and/or object tracking in various examples.

Abstract: Digital fingerprints are used to authenticate a physical item. Authentication attempts may automatically trigger various predetermined processes. Different processes may take place depending on whether the authentication was successful. The process(es) may be computer-implemented, for example, using executable scripts. The triggered process(es) may be immediate (“real-time”), or delayed, deferred, or batched, etc. At least one additional process is triggered or initiated, locally or remotely, by at least one affirmative authentication result or at least one additional process is triggered by a least one unsuccessful authentication attempt. In one example, a combination of purchase and authentication of an item is combined into a single action.

Abstract: In an image processing system, an image processing method, a program, and a recording medium, a person specifying section specifies a plurality of persons appearing in a first composite image owned by a first user. A designated person receiving section receives designation of one or more persons as designated persons among the plurality of persons appearing in the first composite image. An image specifying section specifies images, in which the designated persons appear, from a first image group owned by the first user. A composite image creation unit creates a second composite image using the images in which the designated persons appear.

Abstract: A method for recognizing an action captured on an event-based camera includes the steps of receiving asynchronously transmitted pixel locations which have changed intensity via an address-event bus; stacking a plurality of frames of received pixel location to form a surrogate RGB image where the channels represent individual frames; inputting the surrogate RGB image into a first convolutional neural network; and extracting feature f1 from a last fully connected layer of the convolutional neural network to obtain an action classification of the action, thereby recognizing the action.

Abstract: A dense captioning system and method is provided for analyzing an image to generate proposed bounding regions for a plurality of visual concepts within the image, generating a region feature for each proposed bounding region to generate a plurality of region features of the image, and determining a context feature for the image using a proposed bounding region that is a largest in size of the proposed bounding regions. For each region feature of the plurality of region features of the image, the dense captioning system and method further provides for analyzing the region feature to determine for the region feature a detection score that indicates a likelihood that the region feature comprises an actual object, and generating a caption for a visual concept in the image using the region feature and the context feature when a detection score is above a specified threshold value.

Abstract: Asymmetries are detected in one or more images by partitioning each image to create a set of patches. Salient patches are identified, and an independent displacement for each patch is identified. The techniques used to identify the salient patches and the displacement for each patch are combined in a function to generate a score for each patch. The scores can be used to identify possible asymmetries.

Abstract: Apparatuses and methods associated with preserving a level of confidence of authenticity of an object are disclosed herein. In embodiments, a method includes acquiring first information corresponding to a physical object; identifying first authentication data based on the first information; initializing a database record associated with the physical object responsive to identifying the first authentication data; determining whether the first authentication data corresponds to original provenance for the physical object; in response to determining that the first authentication data does not correspond to original provenance for the physical object, acquire and store in the database record first record data including authenticity information available in association with identification of the first authentication data; in response to determining that the first authentication data does correspond to original provenance for the physical object, retaining second record data in the database record.

Abstract: A local connectivity feature transform (LCFT) is applied to binary document images containing text characters, to generate transformed document images which are then input into a bi-directional Long Short Term Memory (LSTM) neural network to perform character/word recognition. The LCFT transformed image is a gray scale image where the pixel values encode local pixel connectivity information of corresponding pixels in the original binary image. The transform is one that provides a unique transform score for every possible shape represented as a 3×3 block. In one example, the transform is computed using a 3×3 weight matrix that combines bit coding with a zigzag pattern to assign weights to each element of the 3×3 block, and by summing up the weights for the non-zero elements of the 3×3 block shape.

Abstract: A method and system for enhancing spatial resolution of an image of an object includes using an image detector with a sensor array of pixels configured for extended dynamic range detection. Multiple image sample are collected to assemble a stack of image samples and input into an image processor for processing using an image reconstruction engine to generate a single output image from the stack. In a preferred embodiment, the image reconstruction engine executes the Pixon method for image reconstruction.

Abstract: Systems and methods are provided for performing automated face recognition and comparison. An input image of a face may be received and cropped, and the image may be processed through a deep neural network (DNN) to produce a k-dimensional feature vector. The k-dimensional feature vector may be converted to a k-dimensional binary vector by transforming each value in the vector to either 1 or 0. To search for nearest matches of the image in a database of gallery images of faces, the system may compare sub-strings of the binary vector to hash tables created from sub-strings of the gallery images, enabling sub-linear searching that allows locating the closest matches from among the entire gallery without requiring an exhaustive linear search of the entire gallery.

Abstract: Provided is an image processing device including a first rearrangement unit configured to rearrange first divided image data, which is image data corresponding to respective first divided regions obtained by dividing images to be processed represented by processing target image data in the horizontal direction and in the vertical direction, for each of second divided regions, obtained by dividing the images to be processed composed of a plurality of the first divided regions, in an order corresponding to the respective second divided regions, a compression processing unit configured to compress respective pieces of second divided image data, which are image data corresponding to the respective second divided regions, by performing a transform in a predetermined scheme, quantization, and variable length encoding on the data, and a second rearrangement unit configured to rearrange the compressed second divided image data in an order corresponding to all of the images to be processed.

Abstract: One or more embodiments increase the visual appeal and organization of presenting digital content items. In particular, one or more embodiments improve the visual appeal of a presentation of digital content items by emphasizing one or more digital content items within the presentation of digital content items. For instance, one or more digital content items may be emphasized within a graphical user interface including a gallery of digital content items. The emphasized digital content items can be selected based on a score reflecting a particular characteristic of the digital content item and/or a metric of user interest in the digital content item. Digital content items with relatively high scores as compared to other digital content items may be emphasized when presented in conjunction with the other digital content items.

Abstract: A dense captioning system and method is provided for processing an image to produce a feature map of the image, analyzing the feature map to generate proposed bounding boxes for a plurality of visual concepts within the image, analyzing the feature map to determine a plurality of region features of the image, and analyzing the feature map to determine a context feature for the image. For each region feature of the plurality of region features of the image, the dense captioning system further provides for analyzing the region feature to determine a detection score for the region feature, calculating a caption for a bounding box for a visual concept in the image using the region feature and the context feature, and localizing the visual concept by adjusting the bounding box around the visual concept based on the caption to generate an adjusted bounding box for the visual concept.

Abstract: Embodiments are directed to using a hash signature of a rendered DOM object of a website to find similar content and behavior on other websites. Embodiments break a DOM into a large number of data portions (i.e., “shingles”), apply a hashing algorithm to the shingles, select a predetermined number of hashes from the hashed shingles according to a selection criteria to create a hash signature, and compare the hash signature to that of a reference page to determine similarity of website DOM object content. Embodiments can be used to identify phishing websites, defaced websites, spam websites, significant changes in the content of a webpage, copyright infringement, and any other suitable purposes related to the similarity between website DOM object content.

Abstract: A system for performing an operation on data using obfuscated representations of the data is disclosed. Obtaining means are configured to obtain a first obfuscated representation of a first data value and obtain a second obfuscated representation of a second data value. A determining means 102 is configured to determine an obfuscated representation of a third data value, by performing the corresponding operations on the obfuscated representation of the first data value and the obfuscated representation of the second data value. Obfuscating means 101 may be configured to generate the first obfuscated representation based on the first data value and generate the second obfuscated representation based on the second data value. De-obfuscating means 103 may be configured to de-obfuscate the obfuscated representation of the third data value in order to obtain the third data value using a system of equations.

Abstract: A method and apparatus for processing data are provided. The processor includes an input buffer, a data extractor, a multiplier, and an adder. The input buffer receives data and stores the data. The data extractor extracts kernel data corresponding to a kernel in the data from the input buffer. The multiplier multiplies the extracted kernel data by a convolution coefficient. The adder calculates a sum of multiplication results from the multiplier.

Abstract: Systems and methods are provided for managing access to registers. In one embodiment, a system may include a processor and a plurality of registers. The processor and the plurality of registers may be integrated into a single device, or may be in separate devices. The plurality of registers may include a first set of registers that are directly accessible by the processor, and a second set of registers that are not directly accessible by the processor. The second set of registers may, however, be accessed indirectly by the processor via the first set of registers. In one embodiment, the first set of registers may include a register for selecting a register bank from the second set of registers, and a register for selecting a particular address within the register bank, to allow indirect access by the processor to the registers of the second set.

Abstract: Asymmetries are detected in one or more images by partitioning each image to create a set of patches. Salient patches are identified, and an independent displacement for each patch is identified. The techniques used to identify the salient patches and the displacement for each patch are combined in a function to generate a score for each patch. The scores can be used to identify possible asymmetries.

Abstract: Systems and methods are provided for managing access to registers. In one embodiment, a system may include a processor and a plurality of registers. The processor and the plurality of registers may be integrated into a single device, or may be in separate devices. The plurality of registers may include a first set of registers that are directly accessible by the processor, and a second set of registers that are not directly accessible by the processor. The second set of registers may, however, be accessed indirectly by the processor via the first set of registers. In one embodiment, the first set of registers may include a register for selecting a register bank from the second set of registers, and a register for selecting a particular address within the register bank, to allow indirect access by the processor to the registers of the second set.

Abstract: As set forth herein, methods and systems for automated image defect detection. An input image is subjected to skew detection and contaminant detection. Upon detection of skew of and/or contaminants in the image data, correction data, including indicia of the contaminants/skew, correction angles and directions are calculated and displayed on an electronic representation of the document. Automatic correction may then be performed to correct the detected defects, allow the defects to remain, or re-scan the original document to generate new image data without the detected defects.

Abstract: Embodiments are directed to using a hash signature of a rendered DOM object of a website to find similar content and behavior on other websites. Embodiments break a DOM into a large number of data portions (i.e., “shingles”), apply a hashing algorithm to the shingles, select a predetermined number of hashes from the hashed shingles according to a selection criteria to create a hash signature, and compare the hash signature to that of a reference page to determine similarity of website DOM object content. Embodiments can be used to identify phishing websites, defaced websites, spam websites, significant changes in the content of a webpage, copyright infringement, and any other suitable purposes related to the similarity between website DOM object content.

Abstract: Methods, apparatuses, and computer readable storage media are provided for determining a depth measurement of a scene using an optical blur difference between two images of the scene. Each image is captured using an image capture device with different image capture device parameters. A corresponding image patch is identified from each of the captured images, motion blur being present in each of the image patches. A kernel of the motion blur in each of the image patches is determined. The kernel of the motion blur in at least one images patch is used to generate a difference convolution kernel. A selected first image patch is convolved with the generated difference convolution kernel to generate a modified image patch. A depth measurement of the scene is determined from an optical blur difference between the modified image patch and the remaining image patch.

Abstract: Various embodiments of the present invention relate to a method, system and computer program product for detecting and recognizing text in the images captured by cameras and scanners. First, a series of image-processing techniques is applied to detect text regions in the image. Subsequently, the detected text regions pass through different processing stages that reduce blurring and the negative effects of variable lighting. This results in the creation of multiple images that are versions of the same text region. Some of these multiple versions are sent to a character-recognition system. The resulting texts from each of the versions of the image sent to the character-recognition system are then combined to a single result, wherein the single result is detected text.

Abstract: The present invention is to appropriately determine a gesture operation detected to perform data processing according thereto. In the present invention, a CPU judges a processing status at the time of the detection of a gesture operation performed on a touch panel, and after determining its gesture operation type according to the processing status, performs data processing according to the operation type. In this case, the CPU determines and evaluates a gesture operation type for each processing status, and updates values (evaluation accumulation values) of a flick priority pointer and a tap priority pointer. Then, based on these values of the flick priority pointer and the tap priority pointer, the CPU determines one of the gesture operation types (flick operation/tap operation).

Abstract: A first image captured using a first exposure and a second image captured using a second exposure higher than the first exposure are obtained. A third image used as a reference for combining processing is created from the first image. An image, a dynamic range of which is extended, is created by weighted-combining the first and second images using adding coefficients, which are associated with pixel values of the third image in advance.

Abstract: Embodiments are directed to using a hash signature of a rendered DOM object of a website to find similar content and behavior on other websites. Embodiments break a DOM into a large number of data portions (i.e., “shingles”), apply a hashing algorithm to the shingles, select a predetermined number of hashes from the hashed shingles according to a selection criteria to create a hash signature, and compare the hash signature to that of a reference page to determine similarity of website DOM object content. Embodiments can be used to identify phishing websites, defaced websites, spam websites, significant changes in the content of a webpage, copyright infringement, and any other suitable purposes related to the similarity between website DOM object content.

Abstract: An image display unit includes an image display module changing for display images each having image and header information, a touch panel input module receiving inputs, an image reading module reading the image and header information of the images, a group definition determination module determining, based on the header information of a first image of the images that is displayed next in response to an operation performed on the touch panel input module, whether or not a group definition in the header information of the first image is the same as a first predetermined group definition, and an image changing module changing the images to the first image for display in a number of steps predetermined for the first predetermined group definition.

Abstract: A method for transitioning between an image of a first video sequence and an image of a second video sequence, wherein an image of the video sequences corresponds to one sub-region of a respective image of the other video sequence, receiving one image from the first video sequence; providing, from the one image, one pre-transition image; for each one pre-transition image: determining a confidence region within the one pre-transition image based on a set of confidence values, each confidence value representing, for a respective block of the one pre-transition image, the level of reliability of determining one or more blocks of a corresponding image in the second video sequence; computing a transition image corresponding to the second video sequence, the transition image being a function of the determined confidence region of the pre-transition image; and displaying the transition image for a duration upon the reception of the second video sequence.

Abstract: An image processing apparatus includes: a reduced image generation unit configured to obtain a reduced image; a gain data generation unit configured to obtain gain data; and a determination unit configured to determine a control gain required to amplify the input image based on the generated gain data, and letting first gain data be the gain data for an image having a relatively high frequency, and second gain data be the gain data having a relatively low frequency, the determination unit determines the control gain based on a first addition result and a second addition result.

Abstract: System for conveying a stream of information. The system includes an encoding device employing a spatio-temporal coding scheme that omits light, including codes embedded therein that are invisible to a user. A receiver that might be a cell phone camera receives light from the encoding device and computer apparatus is programmed with software to decode the received light to generate the stream of information. The encoding device is preferably a video display.

Abstract: A system and method of symbol matching may include a processor configured to determine which pixels of a first symbol are tangent pixels; for each of the determined tangent pixels, determine whether a second symbol includes a pixel corresponding to the tangent pixel that includes at least one same tangent constraint as that of the tangent pixel; accept the first and second symbols as a match based on shared tangent constraints conditional upon a determination that the second symbol includes for each of at least a subset of the tangent pixels of the first symbol a corresponding pixel that includes the at least one same tangent constraint as that of the tangent pixel; and generate a document including a single symbol that is mapped to the first and second symbols.

Abstract: In techniques for adaptive denoising with internal and external patches, example image patches taken from example images are grouped into partitions of similar patches, and a partition center patch is determined for each of the partitions. An image denoising technique is applied to image patches of a noisy image to generate modified image patches, and a closest partition center patch to each of the modified image patches is determined. The image patches of the noisy image are then classified as either a common patch or a complex patch of the noisy image, where an image patch is classified based on a distance between the corresponding modified image patch and the closest partition center patch. A denoising operator can be applied to an image patch based on the classification, such as applying respective denoising operators to denoise the image patches that are classified as the common patches of the noisy image.

Abstract: A ridge flow based fingerprint image quality determination can be achieved independent of image resolution, can be processed in real-time and includes segmentation, such as fingertip segmentation, therefore providing image quality assessment for individual fingertips within a four finger flat, dual thumb, or whole hand image. A fingerprint quality module receives from one or more scan devices ridge-flow—containing imagery which can then be assessed for one or more of quality, handedness, historical information analysis and the assignment of bounding boxes.

Abstract: An information processing method comprises the steps of: inputting an input information of a multi-dimensional array; calculating an accumulated information value corresponding to a position of each element of the input information; and holding the accumulated information value in a buffer having a size of predetermined bits, wherein in the holding step, when an accumulated information value calculated in the calculating step overflows with respect to the size, a part not more than the predetermined bits of the calculated accumulated information value is held as the accumulated information value.

Abstract: Input information of a multidimensional array is divided into a plurality of divided areas, accumulated information is generated by calculating accumulated values at respective element positions of the input information from a corresponding reference location for each of the plurality of divided areas, and the generated accumulated information is held in a memory for each divided area. Calculation using the accumulated information is executed for a predetermined processing range. The input information is divided into the plurality of divided areas so that two neighboring divided areas have an overlapping area, and the overlapping area has a size at least in which the whole processing range fits.

Abstract: Compact descriptors of digital images are produced by detecting interest points representative of the digital images and selecting out of the interest points key points for producing e.g. local and global compact descriptors of the images. The digital images are decomposed into blocks by computing an energy (variance) for each said block and then subjecting the blocks to culling by rejecting those blocks having an energy failing to pass an energy threshold. The interest points are detected only in the blocks resulting from culling, and the key points for producing the compact descriptors are selected out of the interest points thus detected, possibly by using different selection thresholds for local and global compact descriptors, respectively. The number of key points for producing the compact descriptors may be varied e.g. by adaptively varying the number of the interest points detected in the blocks resulting from culling.

Abstract: A method (100) and system (300) is described for processing video data comprising a plurality of images. The method (100) comprising obtaining (104, 106), for each of the plurality of images, a segmentation in a plurality of regions and a set of keypoints, and tracking (108) at least one region between a first image and a subsequent image resulting in a matched region in the subsequent image taking into account a matching between keypoints in the first image and the subsequent image. The latter results in accurate tracking of regions. Furthermore the method may optionally also perform label propagation taking into account keypoint tracking.

Abstract: An image feature extraction device according to an embodiment includes a gradient image calculator generates intensity gradient data with respect to two different directions based on intensity data of image data; and a gradient count unit calculates a covariance matrix for each partial area obtained by dividing the image data based on the intensity gradient data. The image feature extraction device according to the embodiment further includes a feature data output unit calculates two parameters related to a major axis and a minor axis of an ellipse expressed by the covariance matrix, quantizes a range of the logarithms of the parameters for each of the partial area using a predetermined division number, and outputs a feature vector which contains a value only at a dimension corresponding to the quantized range different from the other dimensions.

Abstract: According to an embodiment, a method for filtering descriptors for visual object recognition is provided. The method includes identifying false positive descriptors having a local match confidence that exceeds a predetermined threshold and a global image match confidence that is less than a second threshold. The method also includes training at least one classifier to discriminate between the false positive descriptors and other descriptors. The method further includes filtering feature point matches using the at least one classifier. According to another embodiment, the filtering step may further include removing one or more feature point matches from a result set. According to a further embodiment, a system for filtering feature point matches for visual object recognition is provided. The system includes a hard false positive identifier, a classifier trainer and a hard false positive filter.

Abstract: In the image processing apparatus, image data is divided into large blocks of a prescribed size and the large blocks are subdivided into small blocks by the dividing unit. The number of isolated points in each large block is then calculated by the large block isolated point calculation unit, and the number of isolated points in each small block is then calculated by the small block isolated point calculation units. It is then determined by the halftone-dot region determination unit whether or not the large block is a halftone-dot region. This determination considers both the number of isolated points in the large block and the number of isolated points in each small block.

Abstract: Discerning small, weak features (“specks”) from imagery can be critical in early-stage cancer detection and other applications. In one aspect, the presence of a speck is judged from a large set of “votes” about whether a point in the imagery has a value above or below its neighbors. Hundreds or thousands or more votes can be gathered from one or more images, to thereby—in the aggregate—tend to confirm or refute the presence of a speck at a particular location. Votes can be based on simple pixel differences, e.g., between a pixel at the center of a 7×7 pixel image excerpt, and each of 48 other pixels in the excerpt. More sophisticated methods can employ analyses of triads, quads, and rings, and kernel-based Markov Random Field frameworks, to roll-up to a final conclusion. A great number of other features and arrangements are also detailed. The technology is particularly illustrated in the context of detecting exoplanets from astronomical imagery of remote star systems.

Abstract: In general, techniques are described for performing a vocabulary-based visual search using multi-resolution feature descriptors. A device may comprise one or more processors configured to perform the techniques. The processors may generate a hierarchically arranged data structure to be used when classifying objects included within a query image based on multi-resolution query feature descriptor extracted from the query image at a first scale space resolution and a second scale space resolution. The hierarchically arranged data structure may represent a first query feature descriptor of the multi-resolution feature descriptor extracted at the first scale space resolution and a second corresponding query feature descriptor of the multi-resolution feature descriptor extracted at the second scale space resolution hierarchically arranged according to the first scale space resolution and the second scale space resolution. The processors may then perform a visual search based on the generated data structure.

Abstract: In general, techniques are described for performing a vocabulary-based visual search using multi-resolution feature descriptors. A device may comprise one or more processors configured to perform the techniques. The one or more processors may to apply a partitioning algorithm to a first subset of target feature descriptors to determine a first classifying data structure to be used when performing a visual search with respect to a query feature descriptor. The one or more processors may then apply the partitioning algorithm to a second subset of the target feature descriptors to determine a second classifying data structure to be used when performing the visual search with respect to the same query feature descriptor.

Abstract: Methods and systems for automating insertion of content into media-based projects are disclosed. In one embodiment, images are uploaded from a user device to a server. The server performs image analysis on the uploaded images and obtains automatic image information. After presentation of the initial project, the user manually interacts with the uploaded images to provide user inputs to the project. The system observes the user interactions to generate manual image information based on the user interactions with the project. Based on the automatic image information and the manual image information, the present invention automatically updates a printable photobook or other media-based project after automatically inserting relevant content into the project. This produces a faster and more seamless user experience, and increases the number of customers purchasing the ultimate product by minimizing friction points to purchase.