Abstract: Examples include detecting objects and determining a set of features for the objects. Examples include receiving a first image input, generating a number of feature maps from the first image input using a number of convolution filters, generating a first number of fully connected layers directly based on the number of feature maps, and detecting a number of objects in the first image and determining a set of features for each object from the first number of fully connected layers.

Abstract: Examples include detecting objects and determining a set of features for the objects. Examples include receiving a first image input, generating a number of feature maps from the first image input using a number of convolution filters, generating a first number of fully connected layers directly based on the number of feature maps, and detecting a number of objects in the first image and determining a set of features for each object from the first number of fully connected layers.

Abstract: Examples include detecting objects and determining a set of features for the objects. Examples include receiving a first image input, generating a number of feature maps from the first image input using a number of convolution filters, generating a first number of fully connected layers directly based on the number of feature maps, and detecting a number of objects in the first image and determining a set of features for each object from the first number of fully connected layers.

Abstract: An example video analytics device can include a memory, a processor executing instructions stored in the memory, an active region detector to identify a plurality of active regions of a plurality of video streams, wherein the plurality of active regions are sections in the plurality of video streams that change from a first frame to a second frame, and a bin packer communicatively coupled with the processor to combine the plurality of active regions to produce a multi-batch of an individual slice, wherein the multi-batch of the individual slice is a batch including the plurality of active regions that is processed at once.

Abstract: Examples include detecting objects and determining a set of features for the objects. Examples include receiving a first image input, generating a number of feature maps from the first image input using a number of convolution filters, generating a first number of fully connected layers directly based on the number of feature maps, and detecting a number of objects in the first image and determining a set of features for each object from the first number of fully connected layers.

Abstract: An example video analytics device can include a memory, a processor executing instructions stored in the memory, an active region detector to identify a plurality of active regions of a plurality of video streams, wherein the plurality of active regions are sections in the plurality of video streams that change from a first frame to a second frame, and a bin packer communicatively coupled with the processor to combine the plurality of active regions to produce a multi-batch of an individual slice, wherein the multi-batch of the individual slice is a batch including the plurality of active regions that is processed at once.

Abstract: The disclosure relates to techniques for automatically generating neurostimulation therapy program groups. The techniques may include receiving rating information and information describing actual therapy effects for a plurality of tested programs, and receiving target therapy data describing target therapy effects. The techniques may include automatically generating plurality of program groups based on the rating information and a comparison of actual effects to the target therapy effects. Actual effects and target therapy effects may be, for example, actual paresthesia areas and target paresthesia areas. The techniques may also include determining whether a sufficient number of programs have been tested to generate a desired number of programs groups and, if a sufficient number have not been tested, automatically generating additional programs based on the tested programs, and automatically generating program groups from the tested and automatically generated programs.

Abstract: Techniques that involve application of one or more rules to a “parent” program to generate a plurality of different “child” programs are described. Each of the rules may define a respective electrode configuration modification, and each child program may be a variation of the parent based on a modification of the electrode configuration of the parent according to one of the rules. The systems or devices may generate further generations of child programs from a previous generation child program using the same one or more rules. The child programs may be provided to a user, so that the user may test the efficacy of the new programs, assisting the user in identifying desirable programs. The child programs may be relatively minor variations of the parent program, and the user may “fine tune” a generally desirable parent program by testing the child programs.

Abstract: The invention is for use with a plural component sequential metering system. In the method of the invention, after a selected number of doses have been dispensed, the system stops and calculates how many of those doses have fallen within a predetermined tolerance of the desired ratio between the two materials. If too many doses fall outside the tolerance, the dose size is decreased. This process is repeated until the appropriate number of doses fall within the desired tolerance.