Systems, computer program products, and techniques for detecting and/or reconstructing objects depicted in digital image data within a three-dimensional space are disclosed. The concepts utilize internal features for detection and reconstruction, avoiding reliance on information derived from location of edges. The inventive concepts provide an improvement over conventional techniques since objects may be detected and/or reconstructed even when edges are obscured or not depicted in the digital image data. In one aspect, detecting a document depicted in a digital image includes: detecting a plurality of identifying features of the document, wherein the plurality of identifying features are located internally with respect to the object; projecting a location of one or more edges of the document based at least in part on the plurality of identifying features; and outputting the projected location of the one or more edges of the document to a display of a computer, and/or a memory.

In an illustrative embodiment, methods and systems for automatically assessing damage vulnerability of a property include accessing digital images of a property parcel having a first structure thereon, classifying features visible in the images, including at least one feature of the first structure and at least one feature present in a neighborhood of the property parcel, to determine at least one characteristic of each feature, determining a spatial relationship between a first structure and each manmade and/or natural feature represented by the classified features, and applying a property loss risk profile, based at least in part on the determined characteristics and the determined spatial relationships, to calculate a risk estimate for the first structure under at least one risk scenario.

Systems and methods are disclosed, including a non-transitory computer readable medium storing computer executable instructions that when executed by a processor cause the processor to identify a first image, a second image, and a third image, the first image overlapping the second image and the third image, the second image overlapping the third image; determine a first connectivity between the first image and the second image; determine a second connectivity between the first image and the third image; determine a third connectivity between the second image and the third image, the second connectivity being less than the first connectivity, the third connectivity being greater than the second connectivity; assign the first image, the second image, and the third image to a cluster based on the first connectivity and the third connectivity; conduct a bundle adjustment process on the cluster of the first image, the second image, and the third image.

A ML model arrangement configured for evaluating motion patterns in a sequence of image data structures is described. The ML model arrangement comprises a first ML model configured for predicting a set of key data elements for each image data structure of the sequence of image data structures, a key data element indicating a respective position of a landmark in the image data structure. The ML model arrangement further comprises at least one second ML model, each second ML model being a ML model configured for evaluating a corresponding specific motion pattern. Each second ML model is configured for determining, based on input data comprising at least one of the key data elements predicted for at least one image data structure or data derived therefrom, class labels for each image data structure, said class labels identifying at least one of: at least one motion phase of the specific motion pattern, at least one evaluation point of the specific motion pattern.

A self-contained, low-cost, low-weight guidance system for vehicles is provided. The guidance system can include an optical camera, a case, a processor, a connection between the processor and an on-board control system, and computer algorithms running on the processor. The guidance system can be integrated with a vehicle control system through “plug and play” functionality or a more open Software Development Kit. The computer algorithms re-create 3D structures as the vehicle travels and continuously updates a 3D model of the environment. The guidance system continuously identifies and tracks terrain, static objects, and dynamic objects through real-time camera images. The guidance system can receive inputs from the camera and the onboard control system. The guidance system can be used to assist vehicle navigation and to avoid possible collisions. The guidance system can communicate with the control system and provide navigational direction to the control system.

Methods, systems, and computer programs are presented for adding new features to a network service. A method includes receiving an image depicting an object of interest. A category set is determined for the object of interest and an image signature is generated for the image. Using the category set and the image signature, the method identifies a set of publications within a publication database and assigns a rank to each publication. The method causes presentation of the ranked list of publications at a computing device from which the image was received.

Embodiments of this application disclose a method for determining camera pose information of a camera of a mobile terminal. The method includes: obtaining a first image, a second image, and a template image, the first image being a previous frame of image of the second image, the first image and the second image being images including a respective instance of the template image captured by the mobile terminal using the camera at a corresponding spatial position; determining a first homography between the template image and the second image; determining a second homography between the first image and the second image; and performing complementary filtering processing on the first homography and the second homography, to obtain camera pose information of the camera, wherein the camera pose information of the camera represents a spatial position of the mobile terminal when the mobile terminal captures the second image using the camera.

One aspect of the invention relates to a fully automatic method for calculating the current, geo-referenced position and alignment of a terrestrial scan-surveying device in situ on the basis of a current panoramic image recorded by the surveying device and at least one stored, geo-referenced 3D scan panoramic image.

A method and an apparatus for training a generative adversarial network (GAN) and a method and an apparatus for processing an image are provided. The method for training the GAN includes: obtaining a fine-grained style label (FGSL) associated with the image and inputting the FGSL and a latent vector into a style-based generator in the GAN; the style-based generator generating an first output image based on the FGSL and the latent vector; the projection discriminator determining whether the first output image matches the image based on the FGSL; and adjusting one or more parameters of the GAN and regenerating, by the style-based generator, a second output image based on the FGSL, the latent vector, and the adjusted GAN in response to determining that the first output image does not match the image based on the FGSL.

Disclosed are an image processing method performed by an electronic device. After a target image for motion transfer and at least one source image corresponding to the target image are acquired, multi-dimensional feature extraction is performed on the source and target images to acquire keypoint feature information of corresponding keypoints in the source and target images, and appearance feature information corresponding to the source image, and the keypoint feature information includes keypoint perspective information. Then, perspective transformation is performed on the keypoints according to the keypoint perspective information to acquire optic flow information of the keypoints. Motion information corresponding to the keypoints is then determined based on the optic flow information and the keypoint feature information, and the motion information and the appearance feature information are fused to acquire a processed image of an object in the source image after transferring motion of an object in the target image.