Skip to content

Commit

Permalink
Update table formatting
Browse files Browse the repository at this point in the history
  • Loading branch information
jayanthkoushik committed Jul 27, 2023
1 parent f7fc91f commit 8e416c5
Show file tree
Hide file tree
Showing 4 changed files with 52 additions and 20 deletions.
Binary file modified paper/main.pdf
Binary file not shown.
35 changes: 25 additions & 10 deletions paper/tables_stats/dataset_splits.md
Original file line number Diff line number Diff line change
@@ -1,10 +1,25 @@
Dataset Min. per-class samples Max. per-class samples
------------ --------------------------------------- -------------------------------------------
$\ \ \many \ \ \medium \ \ \few\ $ $\ \ \ \many \ \ \ \medium \ \ \ \few\ $
<!-- -->
ImageNet‑LT $\ 101 \q 20 \q 5\ $ $1,280 \q 100 \q 19\ $
Places‑LT $\ 103 \q 20 \q 5\ $ $4,980 \q 100 \q 19\ $
CIFAR‑100‑LT $\ 102 \q 20 \q 5\ $ $\ph{,0}500 \q \ph{0}98 \q 19\ $
iNaturalist $\ 101 \q 20 \q 2\ $ $1,000 \q 100 \q 19\ $

: Minimum and maximum per-class training samples for long-tailed datasets. {#tbl:dataset_splits}
\hypertarget{tbl:dataset_splits}{}
\begin{table}
\centering
\begin{tabular}{@{}lrrrcrrr@{}}
\toprule

\multirow{2}*{Dataset} &
\multicolumn{3}{c}{Min. class samples} &&
\multicolumn{3}{c}{Max. class samples} \\

\cmidrule{2-4} \cmidrule{6-8}

& many & med. & few && many & med. & few \\

\midrule

ImageNet‑LT & 101 & 20 & 5 && 1,280 & 100 & 19 \\
Places‑LT & 103 & 20 & 5 && 4,980 & 100 & 19 \\
CIFAR‑100‑LT & 102 & 20 & 5 && 500 & 98 & 19 \\
iNaturalist & 101 & 20 & 2 && 1,000 & 100 & 19 \\

\bottomrule
\end{tabular}
\caption{Minimum and maximum per-class training samples for long-tailed datasets.\label{tbl:dataset_splits}}
\end{table}
36 changes: 26 additions & 10 deletions paper/tables_stats/dataset_stats.md
Original file line number Diff line number Diff line change
@@ -1,10 +1,26 @@
Dataset \ \ \ \ \ Samples \ \ \ \ \ Classes \ \ \ \ \ \ \ \ \ Train samples
------------ ------------------------------------------- ------------------------------------------------------------------ --------------------------------------------------
$\qq \train \qq \val \qq \test$ $\q\ \total \q \many\ \ \q \medium \q \few$ $\qq \many\ \ \q \medium \q \ \ \few$
<!-- -->
ImageNet‑LT $\qq 115,846 \q 20,000 \q 50,000$ $\qq 1,000 \q 385 \q \ph{,0}479 \q 136\ph{,0}$ $\qq \ph{0}88,693 \q 25,510 \q 1,643\ph{0}$
Places‑LT $\qq \ph{0}62,500 \q \ph{0}7,300 \q 36,500$ $\qq 365\ph{,0} \q 131 \q \ph{,0}163 \q 71\ph{0,0}$ $\qq \ph{0}52,762 \q \ph{0}8,934 \q 804\ph{0,0}$
CIFAR‑100‑LT $\qq \ph{0}10,847 \q \ph{00,000} \q 10,000$ $\qq 100\ph{,0} \q \ph{0}35 \q \ph{0,0}35 \q 30\ph{0,0}$ $\qq \ph{00}8,824 \q \ph{0}1,718 \q 305\ph{0,0}$
iNaturalist $\qq 437,513 \q \ph{00,000} \q 24,426$ $\qq 8,142 \q 842 \q 4,076 \q 3,224$ $\qq 258,340 \q 133,061 \q 46,112$

: Statistics of long-tailed datasets. {#tbl:dataset_stats}
\hypertarget{tbl:dataset_stats}{}
\begin{table}
\centering
\begin{tabular}{@{}lrrrclrrrcrrr@{}}
\toprule

\multirow{2}*{Dataset} &
\multicolumn{3}{c}{Samples} &&
\multicolumn{4}{c}{Classes} &&
\multicolumn{3}{c}{Train samples} \\

\cmidrule{2-4} \cmidrule{6-9} \cmidrule{11-13}

& train & val & test && total & many & med. & few && many & med. & few \\

\midrule

ImageNet‑LT & 115,846 & 20,000 & 50,000 && 1,000 & 385 & 479 & 136 && 88,693 & 25,510 & 1,643 \\
Places‑LT & 62,500 & 7,300 & 36,500 && 365 & 131 & 163 & 71 && 52,762 & 8,934 & 804 \\
CIFAR‑100‑LT & 10,847 & - & 10,000 && 100 & 35 & 35 & 30 && 8,824 & 1,718 & 305 \\
iNaturalist & 437,513 & - & 24,426 && 8,142 & 842 & 4,076 & 3,224 && 258,340 & 133,061 & 46,112 \\

\bottomrule
\end{tabular}
\caption{Statistics of long-tailed datasets.\label{tbl:dataset_stats}}
\end{table}
1 change: 1 addition & 0 deletions paper/template.tex
Original file line number Diff line number Diff line change
Expand Up @@ -41,6 +41,7 @@
\usepackage[Export]{adjustbox}
\usepackage{environ}
\usepackage{framed}
\usepackage{multirow}

\setcounter{topnumber}{2}
\setcounter{bottomnumber}{1}
Expand Down

0 comments on commit 8e416c5

Please sign in to comment.