You can use the image generation predict component to generate high-quality images that meet specific characteristics or conditions by using trained models. This component supports various trained GAN models, such as Deep Convolutional GAN (DCGAN), Wasserstein GAN with Gradient Penalty (WGAN-GP), Least Squares GAN (LSGAN), Graph GAN (GGAN), Progressive Growing of GAN (PGGAN), and Style Generative Adversarial Network (StyleGAN). The component allows you to generate new images from random noise. The image generation predict component is widely used in multiple domains, such as image generation, image enhancement, and data enhancement.
Supported computing resources
Inputs and outputs
Input ports
You can use the Read File Data component to read model files from OSS paths.
You can read model files generated by the image generation component.
You can configure the oss path for model parameter of the image generation predict component to select OSS paths in which models reside. For more information, see the parameter descriptions in the following table.
Output port
None.
Configure the component
On the details page of a pipeline in Machine Learning Designer, add the image generation predict component to the pipeline and configure the parameters described in the following table.
Tab | Parameter | Required | Default value | Description | |
Fields Setting | oss path for model | No | None | If no upstream OSS path or model file generated by the image generation component is shipped, you must select the OSS path where the model file resides. | |
oss path for output dir | Yes | None | The OSS path where the output image resides. The output image must be in the same OSS bucket as the pretrained model. | ||
param setting | model type | Yes | dcgan | The image generation network that you want to use. Valid values: DCGAN, WGAN-GP, LSGAN, GGAN, PGGAN, and StyleGAN. | |
num samples | Yes | 16 | The total number of samples to be generated. | ||
number batches | Yes | 4 | The number of samples to be generated in each batch. The samples significantly vary among batches. For example, if 16 samples are to be generated, 4 samples are generated per batch. | ||
Tuning | Select Resource Group | Public Resource Group | No | None | The instance type and virtual private cloud (VPC) that you want to use. You must select a GPU instance type for the algorithm. |
Dedicated resource group | No | None | The number of CPU cores, memory, shared memory, and number of GPUs that you want to use. | ||
Maximum Running Duration (seconds) | No | None | The maximum period of time for which the component can run. If the specified period of time is exceeded, the job is terminated. | ||