Adeia Imaging LLC

Systems and methods are provided for selectively storing and retrieving video data by converting video segments into textual descriptions associated with key visual features and incrementally building a vector database of feature embeddings. The feature embeddings can be used in combination with text-to-video generation models to reconstruct video segments on demand. Selective storage of feature embeddings enables optimized, accurate, and contextually relevant text-to-text and text-to-video generation and video searching functionalities.

Systems and methods for enabling improved content editing at a computing device are disclosed. A first content item is selected, comprising a first plurality of frames and a second plurality of frames that refer to the first plurality of frames. A preferred editing option is identified for the content item, and an icon representing the option is generated for display. Upon receiving user input associated with the icon, the preferred editing option is applied. Applying the option comprises selecting a first frame from the second plurality of frames and a second frame from the first plurality of frames. The selected second frame is utilized as an input to a trained machine learning algorithm to improve the quality of the selected first frame, thereby producing a second content item. This allows for the automated enhancement of lower-quality frames using reference data from high-quality frames.

Systems and methods are provided for improving image item editing. An image item is selected at a computing device and with an editing application, and a preferred editing option to apply to the image item is identified via a user profile. The preferred editing option is determined based on historic editing actions for a plurality of different image items. An icon for applying the preferred editing option to the image item is generated for display in a user interface of the editing application. User input associated with the icon is received, and the preferred editing option is applied to the image item.

Systems and methods are described for automatically performing automated radial blurring. A plurality of images may be accessed, and saliency parameters may be determined based on at least one of the plurality of images. Co-saliency parameters may be determined based on the plurality of images. A region of interest (ROI) in the at least one of the plurality of images may be determined based on the saliency parameters and the co-saliency parameters. Automated radial blurring of the at least one of the plurality of images may be performed based on the identified ROI.

Systems and methods relate to improving image quality of panoramic images. A system generates for display on a device, a user interface for capturing a panoramic image, receives image data from a camera of the device from a plurality of view angles, and generates for display on a device a preliminary panoramic image based on the image data. The system determines that a position corresponding to a particular view angle of the preliminary panoramic image is problematic, and in response to the determining, generates for display an indicator for the preliminary panoramic image prompting for the camera of the device to be repositioned to the particular view angle of the preliminary panoramic image that was determined to be problematic.

Systems, methods and apparatuses are described herein for accessing image data that comprises a plurality of macropixels, wherein the image data may be generated using a device comprising a lenslet array. The image data may be decomposed into a plurality of components using Kronecker product singular value decomposition (KP-SVD). Each component of the plurality of components may be encoded. Each encoded component of the plurality of components may be transmitted to cause display of reconstructed image data based on decoding each encoded component of the plurality of components.

Methods, systems, and devices are disclosed for adjusting an image such that its vector representation more closely aligns with the vector representation of one or more intended search terms, and less closely aligns with the vector representation of one or more non-intended search terms. The method includes accessing an image and the intended and non-intended search terms. The image is iteratively adjusted using a machine learning system operating using a loss function that rewards adjustments resulting in an increase in the similarity score of the intended search terms, and penalizes adjustments resulting in an increase in the similarity score of the non-intended search terms. The loss function also penalizes increases in the perceptual loss between the input image and the adjusted image. The adjusted image may be uploaded to a sharing platform to improve the accuracy of search and organization of the adjusted image.

Systems, methods and apparatuses are described herein for accessing image data that comprises a plurality of macropixels, wherein the image data may be generated using a device comprising a lenslet array. The image data may be decomposed into a plurality of components using Kronecker product singular value decomposition (KP-SVD). Each component of the plurality of components may be encoded. Each encoded component of the plurality of components may be transmitted to cause display of reconstructed image data based on decoding each encoded component of the plurality of components.

Systems and methods for enhancing a live scene to be captured by a camera of a user device based on one or more best matched reference images that include similar subject matter and attributes of the live scene are disclosed. A preview of a live scene is displayed on the user device. Attributes of the live scene are identified. The live scene is segmented separately into a foreground and a background. Vector representations of the attributes of the segmented live scene are calculated and user device parameters are obtained, and both are used to identify reference images that include the attributes of the live scene. The identified reference images are displayed on the user device and a selection is made. The user device parameters are configured based on the selection made to capture the live scene to obtain similar effects of the selected reference image.

Systems and methods relate to improving image quality of panoramic images. A system generates for display on a device, a user interface for capturing a panoramic image, receives image data from a camera of the device from a plurality of view angles, and generates for display on a device a preliminary panoramic image based on the image data. The system determines that a position corresponding to a particular view angle of the preliminary panoramic image is problematic, and in response to the determining, generates for display an indicator for the preliminary panoramic image prompting for the camera of the device to be repositioned to the particular view angle of the preliminary panoramic image that was determined to be problematic.