tu-darmstadt-informatik/bsc-thesis
1
0
Fork 0
mirror of https://github.com/tu-darmstadt-informatik/bsc-thesis.git synced 2026-01-20 20:11:16 +00:00
Code Issues Packages Projects Releases Wiki Activity

WIP

Browse Source
This commit is contained in:
Simon Meister 2017-11-04 19:47:52 +01:00
parent 266dd4179e
commit 523bc5e105
1 changed files with 42 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

44
experiments.tex
View File

@ -97,6 +97,8 @@ is the mean euclidean norm between predicted and ground truth translation, and
E_{p} = \frac{1}{N}\sum_k \lVert p^{gt,i_k} - p^{k,c_k} \rVert
\end{equation}
is the mean euclidean norm between predicted and ground truth pivot.
Analogously, we define error metrics $E_{R}^{cam}$ and $E_{t}^{cam}$ for
predicted camera motion.
\subsection{Training Setup}
Our training schedule is similar to the Mask R-CNN Cityscapes schedule \cite{MaskRCNN}.
@ -108,8 +110,46 @@ first 144K iterations and $0.25 \cdot 10^{-3}$ for all remaining iterations.
\todo{add this}
\subsection{Experiments on Virtual KITTI}
\todo{add this}
\todo{complete this}
{
\begin{table}[t]
\centering
\begin{tabular}{@{}*{10}{c}@{}}
\toprule
\multicolumn{3}{c}{Network} & \multicolumn{3}{c}{Instance Motion Error} & \multicolumn{2}{c}{Camera Motion Error} &\multicolumn{2}{c}{Optical Flow Error} \\
\cmidrule(lr){1-3}\cmidrule(lr){4-6}\cmidrule(l){7-8}\cmidrule(l){9-10}
FPN & cam. & sup. & $E_{R}$ & $E_{t}$ & $E_{p}$ & $E_{R}^{cam}$ & $E_{t}^{cam}$ & AEE & Fl-all \\\midrule
$\times$ & $\times$ & 3D & ? & ? & ? & - & - & ? & ?\% \\
\checkmark & $\times$ & 3D & ? & ? & ? & - & - & ? & ?\% \\
$\times$ & \checkmark & 3D & ? & ? & ? & ? & ? & ? & ?\% \\
\checkmark & \checkmark & 3D & ? & ? & ? & ? & ? & ? & ?\% \\
$\times$ & $\times$ & flow & ? & ? & ? & - & - & ? & ?\% \\
\checkmark & $\times$ & flow & ? & ? & ? & - & - & ? & ?\% \\
$\times$ & \checkmark & flow & ? & ? & ? & ? & ? & ? & ?\% \\
\checkmark & \checkmark & flow & ? & ? & ? & ? & ? & ? & ?\% \\
\bottomrule
\end{tabular}
\caption {
Comparison of network variants on our Virtual KITTI validation set.
AEE: Average Endpoint Error; Fl-all: Ratio of pixels where flow estimate is
wrong by both $\geq 3$ pixels and $\geq 5\%$.
We optionally train camera motion prediction (cam.)
or replace the ResNet50 backbone with ResNet50-FPN (FPN).
We either supervise
object motions (sup.) with 3D motion ground truth (3D) or
with a 2D re-projection loss based on flow ground truth (flow).
Note that for variants where no camera motion is trained and predicted, the optical flow
is composed using the ground truth camera motion and thus the flow error is
only impacted by the predicted 3D object motions.
}
\label{table:vkitti}
\end{table}
}
Table \ref{table:vkitti} compares the performance of different network variants on the Virtual KITTI validation
set.
\subsection{Evaluation on KITTI 2015}
\todo{add this}
\todo{add this if there is enough time}