Monocular 3D Object Detection and Box Fitting Trained End-to-End Using Intersection-over-Union Loss

Monocular 3D Object Detection and Box Fitting Trained End-to-End Using Intersection-over-Union Loss

insights

  • 3DIou loss
  • Homoscedastic noise
    Monocular 3D Object Detection and Box Fitting Trained End-to-End Using Intersection-over-Union Loss_第1张图片
  • Heteroscedastic noise
    Monocular 3D Object Detection and Box Fitting Trained End-to-End Using Intersection-over-Union Loss_第2张图片
  • 3D框的representation
    在这里插入图片描述
    在这里插入图片描述
    在这里插入图片描述
    Monocular 3D Object Detection and Box Fitting Trained End-to-End Using Intersection-over-Union Loss_第3张图片

重要结论

Monocular 3D Object Detection and Box Fitting Trained End-to-End Using Intersection-over-Union Loss_第4张图片

  • distance 越远其variance越大很容易想
  • 而对于图像中的车框,越远车框变化越小,而越近变化越大,所以variance近大远小
    Monocular 3D Object Detection and Box Fitting Trained End-to-End Using Intersection-over-Union Loss_第5张图片

你可能感兴趣的:(deep,learning,3d,bounding,box,3d,bouding,box)