Before issue MLL csv I spent several days writing programs, which
helped me in my work. One of these programs checks the files for
duplicates, the other one finds gaps in numeration. And the last one
checks all JPEG files for errors and compares them to each other. It
does very strict comparison. All the files must have the same number
of JPEG segments in the same order, all corresponding segments must
have equal lengths and the same contents except the entropy-coded data
segments and image sizes, which may be different. I didn't do support
of Exif headers yet, but MLL pictures don't contain them.
So, all files from MLL_CD.csv are in my csv except the following five:
alice133.jpg
belnpat129.jpg
emma246.jpg
viola055.jpg
viola061.jpg
The first three didn't pass my check program because they contained
garbage at the end of Comment JPEG segment. I removed that garbage (2
or 3 bytes were replaced with zeroes) and the files became to be
identical to the other files. The last two files was corrupt (without
EOI marker). I replaced them with good versions.
I also have three rotated pictures which are not authentic. I didn't
include them into csv:
cristina085.jpg
kitty066.jpg
kitty077.jpg
And the last thing I did is renaming some files so all the filenames
have 3-digit numbers (additional leading zeros was added to those
filenames which had 2-digit numbers).
If you have MLL pictures that appear good but are not in my csv,
please post them.
|
|