Research Article
Prediction of Future Appearances via Convolutional Recurrent Neural Networks Based on Image Time Series in Cloud Computing
@INPROCEEDINGS{10.1007/978-3-030-48513-9_27, author={Zao Zhang and Xiaohua Li}, title={Prediction of Future Appearances via Convolutional Recurrent Neural Networks Based on Image Time Series in Cloud Computing}, proceedings={Cloud Computing, Smart Grid and Innovative Frontiers in Telecommunications. 9th EAI International Conference, CloudComp 2019, and 4th EAI International Conference, SmartGIFT 2019, Beijing, China, December 4-5, 2019, and December 21-22, 2019}, proceedings_a={CLOUDCOMP}, year={2020}, month={6}, keywords={Cloud computing CRNN Image series Forecasting}, doi={10.1007/978-3-030-48513-9_27} }- Zao Zhang
Xiaohua Li
Year: 2020
Prediction of Future Appearances via Convolutional Recurrent Neural Networks Based on Image Time Series in Cloud Computing
CLOUDCOMP
Springer
DOI: 10.1007/978-3-030-48513-9_27
Abstract
In recent years, cloud computing has become a prevalent platform to run artificial intelligence (AI) and deep learning applications. With cloud services, AI models can be deployed easily for the convenience of users. However, although cloud service providers such as Amazon Web Services (AWS) have provided various services to support AI applications, the design of AI models is still the key in many specific applications such as forecasting or prediction. For example, how to forecast the future appearance of ornamental plants or pets? To deal with this problem, in this paper we develop a convolutional recurrent neural network (CRNN) model to forecast the future appearance according to their past appearance images. Specifically, we study the problem of using the pine tree’s past appearance images to forecast its future appearance images. We use a plant simulation software to generate pine tree’s growing images to train the model. As a result, our model can generate the future appearance image of the pine tree, and the generated images are very similar to the true images. This means our model can work well to forecast the future appearance based on the image series.
