A method applies a distance-based loss function to a boundary recognition model. The method classifies boundaries of an input with the boundary recognition model. The method also performs semantic segmentation based on the classifying of the boundaries, and outputting a segmentation map showing different classes of objects from the input, based on the semantic segmentation. The method may train an inverse transforming artificial neural network to predict a perspective transformation of an image so that the trained artificial neural network represents the distance-based loss function. The method may freeze weights of the inverse transforming artificial neural network, after training, to obtain the distance-based loss function. Training of the inverse transforming artificial neural network may include generating shifted, translated, and scaled versions of the image such that a ground truth comprises values corresponding to the amounts of shifting, translating, and scaling.

A deep neural network (DNN) can be trained based on a first training dataset that includes first images including annotated first objects. The DNN can be tested based on the first training dataset to determine first object predictions including first uncertainties. The DNN can be tested by inputting a second training dataset and outputting first object predictions including second uncertainties, wherein the second training dataset includes second images including unannotated second objects. A subset of images included in the second training dataset can be selected based on the second uncertainties, The second objects in the selected subset of images included in the second training dataset can be annotated. The DNN can be trained based on the selected subset of images included in the second training dataset including the annotated second objects.

Methods and apparatus for the application of machine learning to radiographic images of animals. In one embodiment, the method includes receiving a set of radiographic images captured of an animal, applying one or more transformations to the set of radiographic images to create a modified set, segmenting the modified set using one or more segmentation artificial intelligence engines to create a set of segmented radiographic images, feeding the set of segmented radiographic images to respective ones of a plurality of classification artificial intelligence engines, outputting results from the plurality of classification artificial intelligence engines for the set of segmented radiographic images to an output decision engine, and adding the set of segmented radiographic images and the output results from the plurality of classification artificial intelligence engines to a training set for one or more of the plurality of classification artificial intelligence engines. Computer-readable apparatus and computing systems are also disclosed.

Embodiments are generally directed to sparse 3D convolution acceleration in a convolutional layer of an artificial neural network model. An embodiment of an apparatus includes one or more processors including a graphics processor to process data; and a memory for storage of data, including feature maps. The one or more processors are to provide for sparse 3D convolution acceleration by applying a shared 3D convolutional kernel/filter to an input feature map to produce an output feature map, including increasing sparsity of the input feature map by partitioning it into multiple disjoint input groups; generation of multiple disjoint output groups corresponding to the input groups by performing a convolution calculation represented by the shared 3D convolutional kernel/filter on all feature values associated with active/valid voxels of each input group to produce corresponding feature values within corresponding output groups; and outputting the output feature map by sequentially stacking the output groups.

According to an example aspect of the present invention, there is provided an apparatus comprising memory configured to store convolutional artificial neural network information comprising at least one filter definition, and at least one processing core configured to generate, from a preceding layer, a convolutional result of a succeeding layer of the artificial neural network in accordance with the at least one filter definition, and generate, from the convolutional result, an activation result of the succeeding layer by using an activation function, the activation function taking three arguments, the three arguments being derived from the convolutional result.

An object recognition unit detects, from image data that represents an image using respective signal values of a plurality of pixels, a first region that is a region representing the object in the image. The object recognition unit determines a first confidence level that is a class confidence level for a first class, the first class being a class of the object represented in the first region. A region recognition unit segments the image of the image data into second regions representing different classes of an object, and determines, for each of second regions, a second class being a class of an object in the second region. An object determination unit determines, as the second class, a class of an object in a non-overlapping region that is the second regions that do not overlap with the first region.

Methods and systems for managing inventory at a retail facility include an image capture device having a field of view that includes a product display shelf of the retail facility, an electronic database that stores a planogram of product display shelves at the retail facility and a computing device. The computing device obtains an image of a product display shelf, detects the individual packages of each of the different products captured in the image, defines different product display areas for the different products on the product display shelf captured in the image, and generates virtual boundary lines that surround each of the defined product display areas. The computing device also determines an identity of a product contained located in each defined product display area, and associates the virtual boundary lines surrounding each defined product display area with an identifier unique to the product contained in the defined product display area.

The subject matter of this specification can be implemented in, among other things, methods, systems, computer-readable storage medium. A method can include receiving, by a processing device, image data including one or more image frames indicative of a current state of a meal preparation area. The processing device determines one of a meal preparation item or a meal preparation action associated with the current state of the kitchen based on the image data. The processing device receives order data comprising one or more pending meal orders. The processing device can determine an order preparation error based on the order data and at least one of the meal preparation item or the meal preparation action. The processing device causes the order preparation error to be displayed on a graphical user interface (GUI).

A moving body, which a person boards, includes an acquisition unit configured to acquire current position information of the moving body, and a recommended path calculation unit configured to calculate a path that is a recommended path toward a predetermined destination and passes through an interior of at least one building, based on the current position information and the predetermined destination.

Performing semantic segmentation in an absence of labels for one or more semantic classes is provided. One or more weak predictors are utilized to obtain label proposals of novel classes for an original dataset for which at least a subset of sematic classes are unlabeled classes. The label proposals are merged with ground truth of the original dataset to generate a merged dataset, the ground truth defining labeled classes of portions of the original dataset. A machine learning model is trained using the merged dataset. The machine learning model is utilized for performing semantic segmentation on image data.