Abstract: Vision Transformers (ViT) have shown their competitive advantages performance-wise compared to convolutional neural networks (CNNs) though they often come with high computational costs. Methods, systems and techniques herein learn instance-dependent attention patterns, utilizing a lightweight connectivity predictor module to estimate a connectivity score of each pair of tokens. Intuitively, two tokens have high connectivity scores if the features are considered relevant either spatially or semantically. As each token only attends to a small number of other tokens, the binarized connectivity masks are often very sparse by nature providing an opportunity to accelerate the network via sparse computations. Equipped with the learned unstructured attention pattern, sparse attention ViT produces a superior Pareto-optimal trade-off between FLOPs and top-1 accuracy on ImageNet compared to token sparsity (48%˜69% FLOPs reduction of MHSA; accuracy drop within 0.4%).

Abstract: There is provided methods and systems for a hair virtual try on. Gray levels from a swatch image and gray levels from a hair portion of an input image are mapped by matching their respective frequencies to establish a map relationship between the swatch image and the hair portion, wherein the gray levels of the swatch image are associated to respective swatch color values. A pixel in the hair portion is colored based on a swatch color value determined using a gray level of the pixel and the map relationship.

Abstract: Methods, apparatus and techniques shape localized objects using a shape model for rendering a reshaped object with an effect such as for a virtual try on (VTO) experience. A VTO can simulate finger/toe nail effects of a nail product or service to nail objects. An example system comprises a nail localization engine including computational circuitry to localize one or more nail objects in an input image of a hand or foot via one or more deep neural networks; a nail shaping engine including computational circuitry for reshaping, via a trained shape model, the one or more nail object as localized by the nail localization engine; and a rendering component including computational circuitry to render an output image simulating a nail product or nail service applied to the one or more nail objects in accordance with the reshaping by the nail shaping engine to provide a virtual try on experience.

Abstract: Techniques are provided for computing systems, methods and computer program products to produce efficient image-to-image translation by adapting unpaired datasets for supervised learning. A first model (a powerful model) may be defined and conditioned using unsupervised learning to produce a synthetic paired dataset from the unpaired dataset, translating images from a first domain to a second domain and images from the second domain to the first domain. The synthetic data generated is useful as ground truths in supervised learning. The first model may be conditioned to overfit the unpaired dataset to enhance the quality of the paired dataset (e.g. the synthetic data generated). A run-time model such as for a target device is trained using the synthetic paired dataset and supervised learning. The run-time model is small and fast to meet the processing resources of the target device (e.g. a personal user device such as a smart phone, tablet, etc.

Abstract: There are provided methods and computing devices using semi-supervised learning to perform end-to-end video object segmentation, tracking respective object(s) from a single-frame annotation of a reference frame through a video sequence of frames. A known deep learning model may be used to annotate the reference frame to provide ground truth locations and masks for each respective object. A current frame is processed to determine current frame object locations, defining object scoremaps as a normalized cross-correlation between encoded object features of the current frame and encoded object features of a previous frame. Scoremaps for each of more than one previous frame may be defined. An Intersection over Union (IoU) function, responsive to the scoremaps, ranks candidate object proposals defined from the reference frame annotation to associate the respective objects to respective locations in the current frame. Pixel-wise overlap may be removed using a merge function responsive to the scoremaps.

Abstract: Provided is a method and system of modifying a digital image rendered on a display screen.

Abstract: Advertising method and system for creating a personalized advertisement, and for recommending a product being advertised. An uploaded facial photograph is received, the facial photograph including a plurality of facial features. At least one facial feature of the plurality of facial features is detected. Personal information may be extracted by analyzing the at least one detected facial feature. A product being advertised may be recommended, wherein the recommended product applies to a treatment related to the extracted personal information.

Abstract: Advertising method and system for creating a personalized advertisement, and for recommending a product being advertised. An uploaded facial photograph is received, the facial photograph including a plurality of facial features. At least one facial feature of the plurality of facial features is detected. Personal information may be extracted by analyzing the at least one detected facial feature. A product being advertised may be recommended, wherein the recommended product applies to a treatment related to the extracted personal information.

Abstract: A method and system for creating and using a virtual dressing room, by superimposing a non-linearly stretchable object image onto a base image in a display screen of a communication device, the images being planar. The method comprises scanning an encoded indicium associated with the object image, accessing and uploading a URL associated with the object image, the object image including a plurality of object image critical points, accessing the base image at the communication device, the base image including a plurality of base image critical points respectively corresponding to the object image critical points, re-mapping the object image via global transformation of coordinates associated with the object image critical points, such that the re-mapped object image critical points coincide with the respective base image critical points, and superimposing the re-mapped object image onto the base image for display at the display screen.