Abstract: Described herein are systems and methods that allow for dense depth map estimation given input images. In one or more embodiments, a neural network model was developed that significantly differs from prior approaches. Embodiments of the deep neural network model comprises more computationally efficient structures and fewer layers but still produces good quality results. Also, in one or more embodiments, the deep neural network model may be specially configured and trained to operate using a hardware accelerator component or components that can speed computation and produce good results, even if lower precision bit representations are used during computation at the hardware accelerator component.

Abstract: The present disclosure provides a living body detecting method and apparatus, a device and a storage medium. The method comprises: regarding a to-be-detected user, respectively obtaining a first picture and a second picture taken with two near infrared cameras and a third picture taken with a visible light camera; generating a depth map according to the first picture and second picture; determining whether the user is a living body according to the depth map and the third picture. The solution of the present disclosure can be applied to improve accuracy of detection results.

Abstract: A prompt method for an eye-protection mode and apparatus, an electronic device and a storage medium are disclosed. The method includes: recognizing a collected face to determine an age status of a user corresponding to the face; acquiring at least two types of parameter information collected by a terminal which is in use by the user, where the at least two types of parameter information are used to represent eye-protection mode-related information; obtaining prompt information for the eye-protection mode according to the age status and the at least two types of parameter information; displaying the prompt information for the eye-protection mode.

Abstract: Embodiments may include a method and apparatus for detecting model reliability. A method may include: acquiring an infrared reflection image and a visible light reflection image of a photographed non-living body; importing the infrared reflection image and the visible light reflection image into a to-be-detected living body detection model, to obtain a detecting result corresponding to the infrared reflection image and a detecting result corresponding to the visible light reflection image; and matching a set concentration level of the infrared reflective coating on the photographed non-living body with the detecting results, and setting a reliability level of the to-be-detected living body detection model based on the matching results.

Abstract: The disclosure provides a method for generating a three-dimensional virtual image. The method includes: obtaining a face image to be processed and a three-dimensional reference model; obtaining a three-dimensional face model, face attribute information and face image information by inputting the face image to be processed into a trained neural network; obtaining a three-dimensional image model by performing a deformation process on the three-dimensional reference model based on the three-dimensional face model; and obtaining a target virtual image by adjusting the three-dimensional image model based on the face attribute information and the face image information.

Abstract: A method and an apparatus for generating information are provided. The method comprises: acquiring a to-be-detected image and facial region information used to indicate a facial region in the to-be-detected image and obtained by previously performing face detection on the to-be-detected image; extracting a facial image from the to-be-detected image based on the facial region information; inputting the facial image into a pre-trained convolutional neural network to obtain image feature information, the convolutional neural network being used to extract an image feature; analyzing the image feature information to determine a probability of each pixel point included in the facial image belonging to a category indicated by each category identifier in a preset category identifier set; and generating information for indicating the category to which the each pixel point belongs based on the probability. This embodiment implements generation of information full of pertinence.

Abstract: The present disclosure discloses a searching method and apparatus. A specific embodiment of the searching method comprises: receiving a search request sent by a client, the search request including a target picture; analyzing the target picture to extract text information and an image characteristic; matching the text information with text information of each preset picture in a preset picture set to determine a first similarity, matching the image characteristic with an image characteristic of the preset picture in response to the first similarity greater than a first preset threshold, and determining whether the preset picture is determined as an identical picture based on matching results; and acquiring associated information of the identical picture, sending the identical picture and the associated information as a search result to the client to be displayed on the client. The present embodiment improves the search efficiency and reduces the labor cost of the search.

Abstract: Embodiments of the disclosure disclose a method, apparatus, server, and storage medium for incorporating a structured entity, wherein the method for incorporating a structured entity can comprise: selecting a candidate entity associated with a to-be-incorporated structured entity from a knowledge graph, determining the to-be-incorporated structured entity being an associated entity based on prior attribute information of a category of the candidate entity and a preset model, merging the associated entity and the candidate entity, and incorporating the associated entity into the knowledge graph. The embodiments can select a candidate entity, and then integrate a preset model using prior knowledge, which can effectively improve the efficiency and accuracy in associating entities, and reduce the amount of calculation, to enable the structured entity to be simply and efficiently incorporated into the knowledge graph.

Abstract: The present disclosure provides a neural network model training method and apparatus, a living body detecting method and apparatus, a device and a storage medium. It is feasible to, with respect to different scenarios, respectively obtain picture samples under the scenarios, each picture sample having a tag, the tag including a living body or a non-living body and a corresponding scenario; and according to obtained picture samples, train to obtain a neural network model whose output is a feature vector which is discriminative for different scenarios. As such, when the living body detection is performed, it is possible to determine the feature vector of the to-be-detected picture according to the neural network model, then determine a picture sample matched with the to-be-detected picture, and regard the tag of the matched picture sample as the tag of the to-be-detected picture, thereby obtaining the detection result of the to-be-detected picture.

Abstract: The disclosure discloses a cartoonlization processing method for an image, and relates to a field of computational vision, image processing, face recognition, deep learning technologies. The method includes: performing skin color recognition on a facial image to be processed to determine a target skin color of a face in the facial image; processing the facial image by utilizing any cartoonizing model in a cartoonizing model set to obtain a reference cartoonized image corresponding to the facial image in a case that the cartoonizing model set does not contain a cartoonizing model corresponding to the target skin color; determining a pixel adjustment parameter based on the target skin color and a reference skin color corresponding to the any cartoonizing model; and adjusting a pixel value of each pixel point in the reference cartoonized image based on the pixel adjustment parameter, to obtain a target cartoonized image corresponding to the facial image.

Abstract: Embodiments of the present disclosure disclose a method and apparatus for recognizing a face. A specific embodiment of the method includes: acquiring at least two facial images of a to-be-recognized face under different illuminations using a near-infrared photographing device; generating at least one difference image based on a brightness difference between each two of the at least two facial images; determining a facial contour image of the to-be-recognized, face based on the at least one difference image; inputting the at least two facial images, the at least one difference image, and the facial contour image into a pre-trained real face prediction value calculation model to obtain a real face prediction value of the to-be-recognized face; and outputting prompt information for indicating successful recognition of a real face, in response to determining the obtained real face prediction value being greater than a preset threshold.

Abstract: Provided are a method for processing multimodal images, an apparatus, a device and a storage medium. Multiple types of vision sensors are disposed in first preset identity recognition scenario.

Abstract: The present disclosure provides a method and apparatus for generating training data for human face recognition, a device and a computer storage medium, wherein the method comprises: inputting accessory-not-worn face images into a generative network, to obtain accessory-worn face images; using the accessory-worn face images as second training data for building the human face recognition model; wherein the generative network is a generative network in a generative adversarial network obtained by pre-training with first training data, the first training data including the accessory-not-worn face images and accessory-worn face images corresponding to a same user identifier. In the present disclosure, the accessory-worn face images obtained in a data augmentation manner greatly expand the amount of training data for building the human face recognition model, and thereby improve the recognition accuracy of the accessory-worn face images.

Abstract: The present disclosure provides a video blending method, apparatus, electronic device and readable storage medium, and relates to computer vision technologies.