左边图右边表,右边是图也同理
\begin{figure}
\begin{minipage}{0.49\textwidth}
\centering
\begin{subfigure}{0.48\textwidth}
\includesvg[width=\linewidth]{d_40_fdiv_plot.svg}
\caption{$d=40$}
\label{fig:mutual_information_diag_gauss_40d}
\end{subfigure}
\hfill
\begin{subfigure}{0.48\textwidth}
\includesvg[width=\linewidth]{d_80_fdiv_plot.svg}
\caption{$d=80$}
\label{fig:mutual_information_diag_gauss_80d}
\end{subfigure}
\vspace{1em}
\begin{subfigure}{0.48\textwidth}
\includesvg[width=\linewidth]{d_120_fdiv_plot.svg}
\caption{$d=120$}
\label{fig:mutual_information_diag_gauss_120d}
\end{subfigure}
\hfill
\begin{subfigure}{0.48\textwidth}
\includesvg[width=\linewidth]{d_160_fdiv_plot.svg}
\caption{$d=160$}
\label{fig:mutual_information_diag_gauss_160d}
\end{subfigure}
\captionof{figure}{\textcolor{red}{Comparison of MI estimates under five different settings as presented in \citep{choi2022density}. OS-DRE accurately estimates MI across all settings, closely aligning with ground truth. }}
\label{fig:mutual_information_diag_gauss}
\vspace{-1mm}
\end{minipage}%
\hspace{2mm}
\begin{minipage}{0.49\textwidth}
\setlength\tabcolsep{3pt}
\renewcommand{\arraystretch}{1.3}
\captionof{table}{Fusion component ablation study. \label{tab:fusion}
Multi-head attention (MHA), sensor positional encoding (SPE) and spatial confidence map (SCM) all improves the performances. Results are reported in AP@0.50/AP@0.70.}
\begin{footnotesize}
\centering
\begin{tabular}{ccc|ccc}
\hline
MHA & SPE & SCM & OPV2V & CoPerception-UAVs & V2X-Sim \\ \hline
& & & 34.96/13.92 & 63.48/44.23 & 51.2/45.7 \\
\checkmark & & & 38.75/13.28 & 63.99/44.46 & 57.3/50.8 \\
\checkmark & \checkmark & & 39.82/16.43 & 64.34/46.86 & 59.1/52.0 \\
\checkmark & \checkmark & \checkmark & \textbf{47.30/19.30} & \textbf{64.83/47.62} & \textbf{59.1/52.2} \\ \hline
\end{tabular}
\end{footnotesize}
\end{minipage}
\vspace{-1mm}
\end{figure}
4万+

被折叠的 条评论
为什么被折叠?



