Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
research:hflow:start [2015/08/12 15:05]
aneufeld
research:hflow:start [2021/03/02 13:28] (current)
Line 1: Line 1:
-====== ​Monocular 3D Reconstruction of Traffic Scenes ​from Optical Flow ======+====== ​Estimating Vehicle Ego-Motion and Piecewise Planar Scene Structure ​from Optical Flow in a Continuous Framework ​====== 
 + 
 +{{:​research:​hflow:​neufeld_gcpr2015.pdf|GCPR 2015 slides}}
  
 We estimate a 3D scene structure and the vehicle egomotion given the forward and backward optical flow fields between two consecutive frames. ​ We estimate a 3D scene structure and the vehicle egomotion given the forward and backward optical flow fields between two consecutive frames. ​
Line 17: Line 19:
 Given rotation $R \in \text{SO}(3)$,​ translation $t \in S(2)$ and 3D point $X$, the point $X'$ relative to the second camera is given by $X' = R^\top (X - t)$. The translation is restricted to the unit sphere, since the translation norm cannot be estimated without additional information. Given rotation $R \in \text{SO}(3)$,​ translation $t \in S(2)$ and 3D point $X$, the point $X'$ relative to the second camera is given by $X' = R^\top (X - t)$. The translation is restricted to the unit sphere, since the translation norm cannot be estimated without additional information.
 Let $\pi$ denote the projection onto the image plane, Let $\pi$ denote the projection onto the image plane,
-$$ \pi\begin{pmatrix} x_1 \\ x_2 \\ x_3 \end{pmatrix} = \frac{1}{x_3} \begin{pmatrix} x_1 \\ x_2 \end{pmatrix}. $$+$$ \pi\begin{pmatrix} x_1 \\ x_2 \\ x_3 \end{pmatrix} = \frac{1}{x_3} \begin{pmatrix} x_1 \\ x_2 \\ x_3 \end{pmatrix}. $$
 Let $z(x)$ denote the depth of pixel $x$ in the image plane, $X = z(x) x$. Depth can be calculated from the plane parameters, $z(x,v) = (v^\top x)^{-1}$. Let $z(x)$ denote the depth of pixel $x$ in the image plane, $X = z(x) x$. Depth can be calculated from the plane parameters, $z(x,v) = (v^\top x)^{-1}$.
 Optical flow is given by Optical flow is given by
Line 59: Line 61:
 ===== Results ===== ===== Results =====
  
-We demonstrate the output of our algorithms on a few examples here, the paper contains a quantitative evaluation. From top to bottom, each example shows the reference frame, estimated depth and estimated ​normals. Depth and depth error (in pixels) are encoded as follows, the scale is derived from ground truth data.+We demonstrate the output of our algorithms on a few examples here, the paper contains a quantitative evaluation. From top to bottom, each example shows the reference frame, estimated depth and normals ​and the flow error as measured in the paper (click for a full size view). Depth and depth error (in pixels) are encoded as follows, the scale is derived from ground truth data.
  
 {{ :​research:​hflow:​depth2.png?​nolink&​200 |}} {{ :​research:​hflow:​depth2.png?​nolink&​200 |}}
Line 69: Line 71:
 The following examples show the first 30 KITTI stereo/flow training sequences. The following examples show the first 30 KITTI stereo/flow training sequences.
  
-<​html><​div align="​center"></​html>​ 
 | {{ :​research:​hflow:​000000_10.png?​direct&​200 |}} {{ :​research:​hflow:​depth_000000.png?​direct&​200 |}} {{ :​research:​hflow:​normals_lab_000000.png?​direct&​200 |}} {{ :​research:​hflow:​err_000000.png?​direct&​200 |}} | {{ :​research:​hflow:​000001_10.png?​direct&​200 |}} {{ :​research:​hflow:​depth_000001.png?​direct&​200 |}} {{ :​research:​hflow:​normals_lab_000001.png?​direct&​200 |}} {{ :​research:​hflow:​err_000001.png?​direct&​200 |}} | {{ :​research:​hflow:​000002_10.png?​direct&​200 |}} {{ :​research:​hflow:​depth_000002.png?​direct&​200 |}} {{ :​research:​hflow:​normals_lab_000002.png?​direct&​200 |}} {{ :​research:​hflow:​err_000002.png?​direct&​200 |}} | | {{ :​research:​hflow:​000000_10.png?​direct&​200 |}} {{ :​research:​hflow:​depth_000000.png?​direct&​200 |}} {{ :​research:​hflow:​normals_lab_000000.png?​direct&​200 |}} {{ :​research:​hflow:​err_000000.png?​direct&​200 |}} | {{ :​research:​hflow:​000001_10.png?​direct&​200 |}} {{ :​research:​hflow:​depth_000001.png?​direct&​200 |}} {{ :​research:​hflow:​normals_lab_000001.png?​direct&​200 |}} {{ :​research:​hflow:​err_000001.png?​direct&​200 |}} | {{ :​research:​hflow:​000002_10.png?​direct&​200 |}} {{ :​research:​hflow:​depth_000002.png?​direct&​200 |}} {{ :​research:​hflow:​normals_lab_000002.png?​direct&​200 |}} {{ :​research:​hflow:​err_000002.png?​direct&​200 |}} |
 | ||| | |||
Line 89: Line 90:
 | ||| | |||
 | {{ :​research:​hflow:​000027_10.png?​direct&​200 |}} {{ :​research:​hflow:​depth_000027.png?​direct&​200 |}} {{ :​research:​hflow:​normals_lab_000027.png?​direct&​200 |}} {{ :​research:​hflow:​err_000027.png?​direct&​200 |}} | {{ :​research:​hflow:​000028_10.png?​direct&​200 |}} {{ :​research:​hflow:​depth_000028.png?​direct&​200 |}} {{ :​research:​hflow:​normals_lab_000028.png?​direct&​200 |}} {{ :​research:​hflow:​err_000028.png?​direct&​200 |}} | {{ :​research:​hflow:​000029_10.png?​direct&​200 |}} {{ :​research:​hflow:​depth_000029.png?​direct&​200 |}} {{ :​research:​hflow:​normals_lab_000029.png?​direct&​200 |}} {{ :​research:​hflow:​err_000029.png?​direct&​200 |}} | | {{ :​research:​hflow:​000027_10.png?​direct&​200 |}} {{ :​research:​hflow:​depth_000027.png?​direct&​200 |}} {{ :​research:​hflow:​normals_lab_000027.png?​direct&​200 |}} {{ :​research:​hflow:​err_000027.png?​direct&​200 |}} | {{ :​research:​hflow:​000028_10.png?​direct&​200 |}} {{ :​research:​hflow:​depth_000028.png?​direct&​200 |}} {{ :​research:​hflow:​normals_lab_000028.png?​direct&​200 |}} {{ :​research:​hflow:​err_000028.png?​direct&​200 |}} | {{ :​research:​hflow:​000029_10.png?​direct&​200 |}} {{ :​research:​hflow:​depth_000029.png?​direct&​200 |}} {{ :​research:​hflow:​normals_lab_000029.png?​direct&​200 |}} {{ :​research:​hflow:​err_000029.png?​direct&​200 |}} |
-<​html></​div></​html>​+
  
 ===== References ===== ===== References =====