Control method of poetry writing system based on k8s cluster

Abstract

The invention belongs to the fields of deep learning and container cloud platforms, and in particular relates to a control method for a system for writing poems based on pictures established based on a K8s cluster, including: 1) using the Kubeadm method and using TensorFlow to build a K8s cluster; Transfer learning and training to obtain a trained image recognition model; 3) Distributed training of the poem generation model through K8s cluster to obtain the trained poem generation model; 4) Use the trained image recognition model to identify the user's uploaded image; 5 ) connect the picture recognition model after the training and the poem generation model after the training by the connection program; 6) adopt the poem generation model after the training to generate a complete poem; 7) show the user all the poems by the output end of the poem The uploaded pictures and the complete poem are described above. The poems generated by the present invention conform to the content of the pictures, and the use of multi-label ancient style category keywords is more consistent with the style of ancient poems.

Description

technical field [0001] The invention belongs to the field of deep learning and container cloud platforms, and in particular relates to a control method for a poetry writing system based on K8s clusters. Background technique [0002] The existing poetry writing system covers incomplete fields, and the experience method is not good. At the same time, the training of the deep learning model is very time-consuming and resource-consuming, and there are many places that can be improved and expanded. For example, "Bian Shi Ji" can write ancient poems or Tibetan acrostic poems based on a given keyword of less than four characters, and "Nine Songs" can write a seven-character quatrain based on a keyword entered. The two systems Both can write ancient poems, but the form of text input does not meet the usage habits of modern users; "Microsoft Xiaobing" can write short modern poems based on pictures uploaded by users, but cannot write ancient poems. Contents of the invention [0003...

AI Technical Summary

Problems solved by technology

[0002] The existing poetry writing system covers incomplete fields, and the experience method is not good. At the same time, the training of the deep learning model is very time-consuming and resource-consuming, and there are many places that can be improved and expanded.

For example, "Bian Shi Ji" can write ancient poems or Tibetan acrostic poems based on a given keyword of less than four characters, and "Nine Songs" can write a seven-character quatrain based on a keyword entered. The two systems Both can write ancient poems, but the form of text input does not conform to the usage habits of modern users; "Microsoft Xiaoice" can write short modern poems based on pictures uploaded by users, but cannot write ancient poems

Images