Abstract: This disclosure describes techniques for performing automatic synthetic shallow depth of field (SDOF) or so-called “portrait mode” suggestions for digital image capture. In particular, these techniques aim to solve the problem of letting users of digital image capture devices know when to turn on portrait image capture modes to capture aesthetically pleasing images. If, based on the application certain criteria, a currently-composed scene is detected to be “portrait worthy,” an icon or other indicator may be provided, e.g., on a user interface of the digital image capture device. According to some embodiments, once a scene has been determined to be portrait worthy, the digital image capture device may capture a subsequent image(s) with at least one additional data asset needed to render a captured image in a synthetic SDOF or portrait mode. This will allow users to convert their digital images into portrait mode images via digital image post-processing operations.

Abstract: Depth perception has become of increased interest in the image community due to the increasing usage of deep neural networks for the generation of dense depth maps. The applications of depth perception estimation, however, may still be limited due to the needs of a large amount of dense ground-truth depth for training. It is contemplated that a self-supervised control strategy may be developed for estimating depth maps using color images and data provided by a sensor system (e.g., sparse LiDAR data). Such a self-supervised control strategy may leverage superpixels (i.e., group of pixels that share common characteristics, for instance, pixel intensity) as local planar regions to regularize surface normal derivatives from estimated depth together with the photometric loss. The control strategy may be operable to produce a dense depth map that does not require a dense ground-truth supervision.

Abstract: Depth perception has become of increased interest in the image community due to the increasing usage of deep neural networks for the generation of dense depth maps. The applications of depth perception estimation, however, may still be limited due to the needs of a large amount of dense ground-truth depth for training. It is contemplated that a self-supervised control strategy may be developed for estimating depth maps using color images and data provided by a sensor system (e.g., sparse LiDAR data). Such a self-supervised control strategy may leverage superpixels (i.e., group of pixels that share common characteristics, for instance, pixel intensity) as local planar regions to regularize surface normal derivatives from estimated depth together with the photometric loss. The control strategy may be operable to produce a dense depth map that does not require a dense ground-truth supervision.

Abstract: A method for computer-aided plate punching is disclosed. The method includes the steps of reading and configuring processing data; reading a design drawing of a workpiece as a processing diagram; selecting needed figures from the processing diagram, and configuring template attributes to confirm contour of the workpiece; selecting operation modes and cutters for processing orifices; processing the orifices and generating cutters information; selecting operation modes and cutters for processing slots; processing the slots and generating cutters information; optimizing the cutters information, and generating a list of cutters; and converting the list of cutters into corresponding CNC codes. A related system is also disclosed.