LVM-Resolution Filter (DLC)

更新时间:
复制 MD 格式

The LVM-Resolution Filter (DLC) component keeps MP4 videos whose resolution falls within a specified range. Use this component before training video generation models to remove videos that are too small or too large for your target resolution.

Supported computing resources

Deep Learning Containers (DLC)

How it works

The component calculates each video's resolution and retains only those that meet the width and height thresholds you define. Videos outside the specified range are excluded from the output.

Inputs and outputs

Input

Connect one of the following as the upstream input:

  • Read File Data: reads the Object Storage Service (OSS) path where your training data is stored. Configure the OSS Data Path parameter to point to the OSS directory containing your video files, or to a video metadata file directly.

  • Any other component in the LVM Data Processing (DLC) series.

Output

The output is written to the OSS directory specified by Output File OSS Path and contains three files:

File Description
{name}.jsonl The filtered results. Pass this file to downstream components as their input. The filename is set by the Output Filename parameter.
{name}_stats.jsonl The state file.
dj_run_yaml.yaml The parameter configuration used during this run, for auditing and reproducing results.

Configure the component

Configure the LVM-Resolution Filter (DLC) component in Machine Learning Designer. The following table describes the parameters.

Tab Parameter Required Description Default
Field Settings Video Data OSS Path No The OSS directory containing your video files, or the path to a video_meta.jsonl metadata file. On the first run without an upstream component, select the OSS directory manually. After the component runs, it generates video_meta.jsonl in the parent directory of the path you specified. On subsequent runs, you can select video_meta.jsonl directly. None
Output File OSS Path Yes The OSS directory where the filtered results are saved. None
Output Filename Yes The filename for the output results file. result.jsonl
Parameter Settings Minimum Width Yes The minimum video width in pixels. Videos narrower than this value are excluded. 1
Maximum Width Yes The maximum video width in pixels. Videos wider than this value are excluded. The default is a recommended starting point — adjust it to match your training target resolution (for example, set to 1920 for 1080p video). 2048
Minimum Height Yes The minimum video height in pixels. Videos shorter than this value are excluded. 1
Maximum Height Yes The maximum video height in pixels. Videos taller than this value are excluded. Adjust to match your training target resolution. 2048
Execution Tuning Number of Processes Yes The number of parallel processes for filtering. Increase this value to speed up processing on large datasets. 4
Select Resource Group Public Resource Group No The instance type (CPU or GPU) and virtual private cloud (VPC) to use. We recommend that you select the CPU instance type to reduce costs. None
Dedicated resource group No The number of vCPUs, memory, shared memory, and GPUs to allocate. None
Maximum Running Duration (seconds) No The time limit for the component run. If the run exceeds this limit, the job is terminated. None