Hello fellow researchers,

I am currently working on a project that involves creating photorealistic 3D environments using smartphone cameras and UE5. I'm looking for insights into quantitative metrics that could effectively evaluate the outcomes of this project, and would appreciate your expert opinions.

  • What are suitable quantitative evaluation metrics for this type of project? I am particularly interested in metrics such as the time taken for environment generation, memory and storage usage, and accuracy of the 3D models.
  • Are there any quantitative methods to assess the realism and visual quality of the generated environments? I'm also curious about methods for quantifying expert evaluations or the quality of visual elements.
  • Additionally, I seek suggestions on performance evaluation metrics for robot simulations conducted in these generated environments. Specifically, I am wondering if evaluation metrics used in object detection systems like "You Only Look Once" (YOLO) are applicable to this project.
  • I would appreciate it if you could let me know if you have any papers that would be helpful.

    I look forward to your valuable insights.

    Best regards,

    Similar questions and discussions