If $E$ is an $n$-dimensional euclidean space then for all $u\in\Hom(E)$ and any orthonormal basis $e_1,\ldots,e_n$: $$ |\det u|\leq\prod_j\norm{u(e_j)}~. $$
$\proof$ 1. For any $w\in\Hom(E)$ we have by the geometric-arithmetic mean inequality: \begin{eqnarray*} \det(w^2)^{1/n} &=&\det(w^*w)^{1/n} =(\l_1\cdots\l_n)^{1/n} \leq\tfrac1n(\l_1+\cdots+\l_n)\\ &=&\tfrac1n\sum\la w^*w(x_j),x_j\ra =\tfrac1n\sum\Vert w(x_j)\Vert^2 =\tfrac1n\Vert w\Vert_{HS}^2 \end{eqnarray*} where $x_1,\ldots,x_n$ is an orthonormal basis of eigen vectors of $w^*w$ with eigen values $\l_j\geq0$.
2. Define $v\in\Hom(E)$ by $v:e_j\mapsto d_je_j$, then for any $u\in\Hom(E)$: $$ (\det v^2\det(uu^*))^{1/n} =(\det(uv)^2)^{1/n} \leq\tfrac1n\norm{(uv)^2}_{HS} =\tfrac1n\sum_j\norm{uv(e_j)}^2~. $$ Now choose $d_j > 0$ such that $\prod d_j=1$ and $d_j\norm{u(e_j)}=d_k\norm{u(e_k)}$; by the equality case in the geometric-arithmetic mean inequality it follows that: $$ \tfrac1n\sum_j\norm{uv(e_j)}^2 =\tfrac1n\sum_jd_j^2\norm{u(e_j)}^2 =\Big(\prod_j d_j\norm{u(e_j)}^2\Big)^{1/n} =\Big(\prod_j \norm{u(e_j)}^2\Big)^{1/n}~. $$ $\eofproof$