A method for training a neural network includes receiving a plurality of images and, for each individual image of the plurality of images, generating a training triplet including a subset of the individual image, a subset of a transformed image, and a homography based on the subset of the individual image and the subset of the transformed image. The method also includes, for each individual image, generating, by the neural network, an estimated homography based on the subset of the individual image and the subset of the transformed image, comparing the estimated homography to the homography, and modifying the neural network based on the comparison.

Embodiments of the present disclosure provide a software program that displays both a volume as images and segmentation results as surface models in 3D. Multiple 2D slices are extracted from the 3D volume. The 2D slices may be interactively rotated by the user to best follow an oblique structure. The 2D slices can “cut” the surface models from the segmentation so that only half of the models are displayed. The border curves resulting from the cuts are displayed in the 2D slices. The user may click a point on the surface model to designate a landmark point. The corresponding location of the point is highlighted in the 2D slices. A 2D slice can be reoriented such that the line lies in the slice. The user can then further evaluate or refine the landmark points based on both surface and image information.

Embodiments of the present disclosure provide a software program that displays both a volume as images and segmentation results as surface models in 3D. Multiple 2D slices are extracted from the 3D volume. The 2D slices may be interactively rotated by the user to best follow an oblique structure. The 2D slices can “cut” the surface models from the segmentation so that only half of the models are displayed. The border curves resulting from the cuts are displayed in the 2D slices. The user may click a point on the surface model to designate a landmark point. The corresponding location of the point is highlighted in the 2D slices. A 2D slice can be reoriented such that the line lies in the slice. The user can then further evaluate or refine the landmark points based on both surface and image information.

Embodiments of the present application disclose time division display control methods and apparatus and display devices, wherein a time division display control method disclosed comprises: displaying a first image by a display device; and changing display pixel distribution of the display device at least once within a preset permitted staying duration of vision of human eyes and displaying the first image by the display device changed each time, to cause the displayed first images to be displayed in human eyes as a second image in an overlapped manner. According to the present application, utilization of display pixels of a display device and a display quality of at least a local part of an image can be improved, thereby better meeting diversified actual application demands of users.

Embodiments of the present application disclose time division display control methods and apparatus and display devices, wherein a time division display control method disclosed comprises: displaying a first image by a display device; and changing display pixel distribution of the display device at least once within a preset permitted staying duration of vision of human eyes and displaying the first image by the display device changed each time, to cause the displayed first images to be displayed in human eyes as a second image in an overlapped manner. According to the present application, utilization of display pixels of a display device and a display quality of at least a local part of an image can be improved, thereby better meeting diversified actual application demands of users.

The present disclosure provides a method of medical procedure using augmented reality for superimposing a patient's medical images (e.g., CT or MRI) over a real-time camera view of the patient. Prior to the medical procedure, the patient's medical images are processed to generate a 3D model that represents a skin contour of the patient's body. The 3D model is further processed to generate a skin marker that comprises only selected portions of the 3D model. At the time of the medical procedure, 3D images of the patient's body are captured using a camera, which are then registered with the skin marker. Then, the patient's medical images can be superimposed over the real-time camera view that is presented to the person performing the medical procedure.

The present disclosure provides a method of medical procedure using augmented reality for superimposing a patient's medical images (e.g., CT or MRI) over a real-time camera view of the patient. Prior to the medical procedure, the patient's medical images are processed to generate a 3D model that represents a skin contour of the patient's body. The 3D model is further processed to generate a skin marker that comprises only selected portions of the 3D model. At the time of the medical procedure, 3D images of the patient's body are captured using a camera, which are then registered with the skin marker. Then, the patient's medical images can be superimposed over the real-time camera view that is presented to the person performing the medical procedure.

The present application discloses a face super-resolution realization method and apparatus, an electronic device and a storage medium, and relate to fields of face image processing and deep learning. The specific implementation solution is as follows: a face part in a first image is extracted; the face part is input into a pre-trained face super-resolution model to obtain a super-sharp face image; a semantic segmentation image corresponding to the super-sharp face image is acquired; and the face part in the first image is replaced with the super-sharp face image, by utilizing the semantic segmentation image, to obtain a face super-resolution image.

The present application discloses a face super-resolution realization method and apparatus, an electronic device and a storage medium, and relate to fields of face image processing and deep learning. The specific implementation solution is as follows: a face part in a first image is extracted; the face part is input into a pre-trained face super-resolution model to obtain a super-sharp face image; a semantic segmentation image corresponding to the super-sharp face image is acquired; and the face part in the first image is replaced with the super-sharp face image, by utilizing the semantic segmentation image, to obtain a face super-resolution image.

Drone-based inventory management method and systems. One embodiment provides a drone-based inventory management system including one or more unmanned aerial vehicles (UAVs), and a central management system having an electronic processor, and a transceiver configured to communicate with the one or more UAVs. The electronic processor is configured to determine a discrepancy in inventory and select a UAV for verification. The electronic processor is also configured to determine whether weather permits UAV operation and operate the UAV in a pre-determined route when the weather permits UAV operation. The electronic processor is further configured to capture images using the UAV and determine new inventory based on captured images. The electronic processor is also configured to update inventory based on the new inventory.