Create an evaluation task

Last updated:2020-11-27 10:35:16

Create an evaluation task

Task information

  • Evaluation Task Name: The name of the task. The task name can contain at most 64 characters, including Chinese characters, letters, digits, spaces, underscores (_), hyphens (-), and periods (.). The task name must be unique.
  • Evaluation Sample Type: The type of the samples to be evaluated. Valid values: Image and Video.
  • Evaluation Mode: The evaluation method of the evaluation task. Valid values: Full Reference Evaluation (Original Image VS Result Set), Full Reference Evaluation (Result Set VS Result Set), No Reference Evaluation, and OAA mode.
  • Original Image or Original Video: The original images or videos that have not been processed.
  • Result Set: The images or videos that are processed. The system can automatically generate a result set based on a model,. Alternatively, you can upload local images or videos that have been processed.
  • Evaluation Dimension: The name of the evaluation standard. You can select a preset evaluation standard or a custom evaluation standard.

Specify the task information

Evaluation methods

  • Image
    1. Full Reference Evaluation (Original Image VS Result Set) or Full Reference Evaluation (Result Set VS Result Set)
    2. No Reference Evaluation
  • Video
    1. Full Reference Evaluation (Original Image VS Result Set) or Full Reference Evaluation (Result Set VS Result Set)
    2. No Reference Evaluation

Algorithm models

KQoE provides independently developed algorithm models to perform super-resolution and smart high-definition processing on original images or videos, improving the image and video quality while saving bandwidth.

Step Description
1 Click New Evaluate Task.
2 Upload original images or videos.
3 Select Model Generate for Result Set.
4 Select an algorithm as required.
5 Click Submit. After the result set is generated, view the evaluation task.

Algorithm models for images

Available algorithm models

The following algorithm models are available for evaluating images:

  1. Face restoration (generated)
  2. Face restoration (denoised)
  3. Face restoration (equalized)

Did you find the above information helpful?

Unhelpful
Mostly Unhelpful
A little helpful
Helpful
Very helpful

What might be the problems?

Insufficient
Outdated
Unclear or awkward
Redundant or clumsy
Lack of context for the complex system or functionality

More suggestions

0/200

Please give us your feedback.

Submitted

Thank you for your feedback.

问题反馈