A method for determining an abnormality in a medical device from a medical image is provided. The method for determining an abnormality in a medical device comprises receiving a medical image, and detecting information on at least a part of a target medical device included in the received medical image.

A point cloud registration method, apparatus, device, and storage medium are provided. The method includes: acquiring target point cloud data; dividing the target point cloud data into a plurality of point cloud sets; determining a coincidence degree between every two point cloud sets and determining a fixed point cloud set and a registration point cloud set from two point cloud sets with a coincidence degree between the two point cloud sets being greater than a preset threshold; determining a target registration matrix between the fixed point cloud set and the registration point cloud set; and performing registration of the fixed point cloud set with the registration point cloud set according to the target registration matrix.

A point cloud registration method, apparatus, device, and storage medium are provided. The method includes: acquiring target point cloud data; dividing the target point cloud data into a plurality of point cloud sets; determining a coincidence degree between every two point cloud sets and determining a fixed point cloud set and a registration point cloud set from two point cloud sets with a coincidence degree between the two point cloud sets being greater than a preset threshold; determining a target registration matrix between the fixed point cloud set and the registration point cloud set; and performing registration of the fixed point cloud set with the registration point cloud set according to the target registration matrix.

A neural network pruning system can sparsely prune neural network models using an optimizer based approach that is agnostic to the model architecture being pruned. The neural network pruning system can prune by operating on the parameter vector of the full model and the gradient vector of the loss function with respect to the model parameters. The neural network pruning system can iteratively update parameters based on the gradients, while zeroing out as many parameters as possible based a preconfigured penalty.

A neural network pruning system can sparsely prune neural network models using an optimizer based approach that is agnostic to the model architecture being pruned. The neural network pruning system can prune by operating on the parameter vector of the full model and the gradient vector of the loss function with respect to the model parameters. The neural network pruning system can iteratively update parameters based on the gradients, while zeroing out as many parameters as possible based a preconfigured penalty.

A global and local binary pattern image crack segmentation method based on robot vision comprises the following steps: enhancing a contrast of an acquired original image to obtain an enhanced map; using an improved local binary pattern detection algorithm to process the enhanced map and construct a saliency map; using the enhanced map and the saliency map to segment cracks and obtaining a global and local binary pattern automatic crack segmentation method; and evaluating performance of the obtained global and local binary pattern automatic crack segmentation method. The present application uses logarithmic transformation to enhance the contrast of a crack image, so that information of dark parts of the cracks is richer. Texture features of a rotation invariant local binary pattern are improved. Global information of four directions is integrated, and the law of universal gravitation and gray and roundness features are introduced to correct crack segmentation results, thereby improving segmentation accuracy. Crack regions can be segmented in the background of uneven illumination and complex textures. The method has good robustness and meets requirements of online detection.

A global and local binary pattern image crack segmentation method based on robot vision comprises the following steps: enhancing a contrast of an acquired original image to obtain an enhanced map; using an improved local binary pattern detection algorithm to process the enhanced map and construct a saliency map; using the enhanced map and the saliency map to segment cracks and obtaining a global and local binary pattern automatic crack segmentation method; and evaluating performance of the obtained global and local binary pattern automatic crack segmentation method. The present application uses logarithmic transformation to enhance the contrast of a crack image, so that information of dark parts of the cracks is richer. Texture features of a rotation invariant local binary pattern are improved. Global information of four directions is integrated, and the law of universal gravitation and gray and roundness features are introduced to correct crack segmentation results, thereby improving segmentation accuracy. Crack regions can be segmented in the background of uneven illumination and complex textures. The method has good robustness and meets requirements of online detection.

A method, a system and a computer readable medium for automatic segmentation of a 3D medical image, the 3D medical image comprising an object to be segmented, the method characterized by comprising: carrying out, by using a machine learning model, in at least two of a first, a second and a third orthogonal orientation, 2D segmentations for the object in slices of the 3D medical image to derive 2D segmentation data; determining a location of a bounding box (10) within the 3D medical image based on the 2D segmentation data, the bounding box (10) having predetermined dimensions; and carrying out a 3D segmentation for the object in the part of the 3D medical image corresponding to the bounding box (10).

A method, a system and a computer readable medium for automatic segmentation of a 3D medical image, the 3D medical image comprising an object to be segmented, the method characterized by comprising: carrying out, by using a machine learning model, in at least two of a first, a second and a third orthogonal orientation, 2D segmentations for the object in slices of the 3D medical image to derive 2D segmentation data; determining a location of a bounding box (10) within the 3D medical image based on the 2D segmentation data, the bounding box (10) having predetermined dimensions; and carrying out a 3D segmentation for the object in the part of the 3D medical image corresponding to the bounding box (10).

The album creation application of the present disclosure displays image data, to which trimming is performed, and a template, which includes a slot in which the image data is arranged, so that a slot and image data to be arranged in the slot are selected by use of an input device. Position information of a point of interest in the image to be arranged in the slot is obtained. Composition patterns applicable to the image with designation of the point of interest are presented to the user, and the composition pattern to be applied, which is selected by the user from among the presented composition patterns, is obtained. Trimming is performed based on the point of interest and the selected composition pattern selected. The trimmed images are listed, so that multiple trimmed images are presented to the user as trimming proposals.