Provisioning infrastructure from visual diagrams
Provisioning infrastructure from visual diagrams can include selecting, with a first neural network, regions of interest within an infrastructure architecture diagram that corresponds to an infrastructure offered by an infrastructure provider for supporting a computer application. With a second neural network, infrastructure resources and interconnections among the infrastructure resources can be identified based on objects appearing within the regions of interest. Properties corresponding to the infrastructure resources identified can be determined with a third neural network. With a fourth neural network, an infrastructure architecture specification can be generated, the infrastructure architecture specification specifying the infrastructure resources, corresponding properties of the infrastructure resources, and interconnections among the infrastructure resources. The infrastructure architecture specification can be used to configure networked resources to support the computer application.
1. A method, comprising:
selecting by a first neural network, using computer hardware, regions of interest within an infrastructure architecture diagram corresponding to an infrastructure offered by an infrastructure provider to support a computer application;
identifying by a second neural network, using the computer hardware, infrastructure resources and interconnections among the infrastructure resources based on objects appearing within the regions of interest;
determining by a third neural network, using the computer hardware, properties corresponding to the infrastructure resources identified; and
generating by a fourth neural network, using the computer hardware, an infrastructure architecture specification specifying the infrastructure resources, corresponding properties of the infrastructure resources, and interconnections among the infrastructure resources, wherein the infrastructure architecture specification is usable for configuring networked resources to support the computer application.
2. The method of claim 1 , wherein the selecting comprises:
generating images of each of the regions of interest; and
grouping the images into a first group and a second group, wherein each image of the first group contains a single object, and wherein each image of the second group contains an interconnected pair of objects.
3. The method of claim 2 , wherein the first neural network implements a single shot multi-box detector (SSD) model, and wherein each image corresponds to a bounding box generated based on the SSD model.
4. The method of claim 1 , wherein each object comprises one of a plurality of graphical symbols, each of which is predetermined to correspond to only one of a plurality of infrastructure resources, and wherein the second neural network is trained using machine learning to identify each infrastructure resource based on a corresponding graphical symbol.
5. The method of claim 1 , further comprising automatically configuring networked resources to support the application according to the infrastructure architecture specification.
6. The method of claim 1 , wherein the generating further comprises presenting the infrastructure architecture specification to a user and modifying infrastructure architecture specification in response to user-supplied input and submitting the infrastructure architecture specification as modified to the infrastructure provider.
7. The method of claim 1 , further comprising generating feedback based on user-supplied input and updating at least one of the first neural network, the second neural network, the third neural network, and the fourth neural network based on the feedback.
8. A system, comprising:
a processor configured to initiate operations including:
selecting, with a first neural network, regions of interest within an infrastructure architecture diagram corresponding to an infrastructure offered by an infrastructure provider to support a computer application;
identifying, with a second neural network, infrastructure resources and interconnections among the infrastructure resources based on objects appearing within the regions of interest;
determining, with a third neural network, properties corresponding to the infrastructure resources identified; and
generating, with a fourth neural network, an infrastructure architecture specification specifying the infrastructure resources, corresponding properties of the infrastructure resources, and interconnections among the infrastructure resources, wherein the infrastructure architecture specification is usable for configuring networked resources to support the computer application.
9. The system of claim 8 , wherein the selecting includes:
generating images of each of the regions of interest; and
grouping the images into a first group and a second group, wherein each image of the first group contains a single object, and wherein each image of the second group contains an interconnected pair of objects.
10. The system of claim 9 , wherein the first neural network implements a single shot multi-box detector (SSD) model, and wherein each image corresponds to a bounding box generated based on the SSD model.
11. The system of claim 8 , wherein each object comprises one of a plurality of graphical symbols, each of which is predetermined to correspond to only one of a plurality of infrastructure resources, and wherein the second neural network is trained using machine learning to identify each infrastructure resource based on a corresponding graphical symbol.
12. The system of claim 8 , wherein the processor is configured to initiate operations further including automatically configuring networked resources to support the application according to the infrastructure architecture specification.
13. The system of claim 8 , wherein the generating further includes presenting the infrastructure architecture specification to a user and modifying infrastructure architecture specification in response to user-supplied input and submitting the infrastructure architecture specification as modified to the infrastructure provider.
14. A computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to initiate operations comprising:
selecting with a first neural network, by the processor, regions of interest within an infrastructure architecture diagram corresponding to an infrastructure offered by an infrastructure provider to support a computer application;
identifying with a second neural network, by the processor, infrastructure resources and interconnections among the infrastructure resources based on objects appearing within the regions of interest;
determining with a third neural network, by the processor, properties corresponding to the infrastructure resources identified; and
generating with a fourth neural network, by the processor, an infrastructure architecture specification specifying the infrastructure resources, corresponding properties of the infrastructure resources, and interconnections among the infrastructure resources, wherein the infrastructure architecture specification is usable for configuring networked resources to support the computer application.
15. The computer program product of claim 14 , wherein the selecting comprises:
generating images of each of the regions of interest; and
grouping the images into a first group and a second group, wherein each image of the first group contains a single object, and wherein each image of the second group contains an interconnected pair of objects.
16. The computer program product of claim 15 , wherein the first neural network implements a single shot multi-box detector (SSD) model, and wherein each image corresponds to a bounding box generated based on the SSD model.
17. The computer program product of claim 14 , wherein each object comprises one of a plurality of graphical symbols, each of which is predetermined to correspond to only one of a plurality of infrastructure resources, and wherein the second neural network is trained using machine learning to identify each infrastructure resource based on a corresponding graphical symbol.
18. The computer program product of claim 14 , wherein the program instructions are executable by the processor to cause the processor to initiate operations further comprising automatically configuring networked resources to support the application according to the infrastructure architecture specification.
19. The computer program product of claim 14 , wherein the generating further comprises presenting the infrastructure architecture specification to a user and modifying infrastructure architecture specification in response to user-supplied input and submitting the infrastructure architecture specification as modified to the infrastructure provider.
20. The computer program product of claim 14 , wherein the program instructions are executable by the processor to cause the processor to initiate operations further comprising generating feedback based on user-supplied input and updating at least one of the first neural network, the second neural network, the third neural network, and the fourth neural network based on the feedback.