The LVM data processing algorithm provides capabilities for video cleaning, video classification, content filtering, basic information extraction, and caption generation. You can combine algorithms to filter video data and generate text descriptions. This prepares high-quality video data for training video generation models. This topic describes how to use the Video Data Filtering and Labeling preset template in Visualized Modeling (Designer).
Limitations
The Video Data Filtering and Labeling preset template is available only in the following regions: China (Hangzhou), China (Shanghai), China (Beijing), and China (Shenzhen).
Prepare video data
Upload your video files to Object Storage Service (OSS). The video files must meet the following requirements:
-
The video files must be in MP4 format.
-
The file names must not contain Chinese characters.
Create and run a pipeline
-
Go to the Visualized Modeling (Designer) page.
-
Log on to the PAI console.
-
In the upper-left corner, select a region as needed.
-
In the left-side navigation pane, click Workspaces, and then click the name of your workspace to open it.
-
In the left-side navigation pane, choose Model Training > Visualized Modeling (Designer).
-
-
Create a pipeline.
-
On the Preset Templates tab, choose Business Area > multi-modal large model, and then click Create on the Video Data Filtering and Labeling card.
-
Configure the pipeline parameters, or keep the default settings, and then click Confirm.
-
From the pipeline list, select the pipeline that you created and click Open.
-
-
Configure the pipeline.
The key configurations are as follows:
-
Read OSS Data component: Set OSS Data Path to the OSS directory where you uploaded the video files in the Prepare video data step.
-
LVM algorithm components: Configure each LVM algorithm component. For details, see Video preprocessing operators.
-
-
Run the pipeline. After the pipeline is complete, view the generated files:
-
video_meta.jsonl file: The pipeline generates a metadata file named video_meta.jsonl in the parent directory of the path specified for Video Data OSS Path.
-
Result file: View the result file in the path specified for Output File OSS Path.
For more information about the result file, see the description of the Output File OSS Path parameter in Video preprocessing operators.
-