Not quite. If I have one of the OP's input lines:

Code:

 D|123457|Jack Roes|1/5/2016|1/5/2016|LR|LW|||||||||||||||||

the output I get is:

Code:

 D|123457|Jack Roes|01/5/2016|01/5/2016|LR|LW||||||||||||||||||

This is because the first match on the BRE is |1/ and you want to match /5/ to change the 5 to 05, but the leading / for the 2nd match was consumed as the trailing / in the previous match.

This can be fixed by applying that substitute command twice:

Code:

sed -e 's/\([^0-9]\)\([0-9]\/\)/\10\2/g' -e 's/\([^0-9]\)\([0-9]\/\)/\10\2/g' /path/to/your/file

or with a slightly simpler second substitute command:

Code:

sed -e 'sx\([^0-9]\)\([0-9]/\)x\10\2xg' -e 'sx/\([0-9]/\)x/0\1xg' /path/to/your/file

(using x as the substitution delimiter instead of / to avoid a few backslashes).

Note that this sed solution will also work if a date field also includes time (e.g., 1/5/2016 11:23:45) while the awk solutions suggested will fail in this case. And note that this sed solution may "fix" too much if there is a field containing a simple fraction where the numerator is a single digit (e.g., 1/2 off will become 01/2 off and 2/3 majority will become 02/3 majority). None of these are problems with the given sample input, but with all of those empty fields in the sample we have to wonder what else might appear in those field in other input lines???