I would like to produce the following:

      a       b
    xxxxx   xxxxx
 1  xxxxx   xxxxx
    xxxxx   xxxxx

    xxxxx   xxxxx
 2  xxxxx   xxxxx
    xxxxx   xxxxx

Where the blocks of 'x's are images, and 'a', 'b', '1' and '2' are text.

Here are my two attempts so far:

\begin{figure}
\begin{center}
\begin{tabular}{ccc}
 & a & b \\
1 & \subfloat[]{\includegraphics[width=0.47\textwidth]{im.png}} &
    \subfloat[]{\includegraphics[width=0.47\textwidth]{im.png}} \\
2 & \subfloat[]{\includegraphics[width=0.47\textwidth]{im.png}} &
    \subfloat[]{\includegraphics[width=0.47\textwidth]{im.png}} \\
\end{tabular}
\end{center}
\end{figure}

Which produces:

      a       b
    xxxxx   xxxxx
    xxxxx   xxxxx
 1  xxxxx   xxxxx

    xxxxx   xxxxx
    xxxxx   xxxxx
 2  xxxxx   xxxxx

And

\begin{figure}
\begin{center}
\begin{tabular}{m{1cm}m{6cm}m{6cm}}
 & a & b \\
1 & \subfloat[]{\includegraphics[width=0.47\textwidth]{im.png}} &
    \subfloat[]{\includegraphics[width=0.47\textwidth]{im.png}} \\
2 & \subfloat[]{\includegraphics[width=0.47\textwidth]{im.png}} &
    \subfloat[]{\includegraphics[width=0.47\textwidth]{im.png}} \\
\end{tabular}
\end{center}
\end{figure}

Which produces:

    a       b
    xxxxx   xxxxx
 1  xxxxx   xxxxx
    xxxxx   xxxxx

    xxxxx   xxxxx
 2  xxxxx   xxxxx
    xxxxx   xxxxx