Add dataset download script

download.sh

@@ -0,0 +1,37 @@
+#!/usr/bin/env bash
+##
+## Download the fMoW-RGB-baseline dataset (train and validation data) in the specified directory
+## Requires `wget`, `tar` and `pv` (optional)
+##
+set -eu
+
+if [ $# -ne 1 ]; then
+    echo "Usage: $0 <target_directory>"
+    exit 1
+fi
+
+target_dir=$1
+mkdir -p "${target_dir}" && cd "${target_dir}"
+
+repository_url="https://huggingface.co/datasets/jbourcier/fmow-rgb-baseline"
+
+echo "Downloading archives..."
+for archive in 'train-images.tar.gz' 'train-metadata.tar.gz' 'val-images.tar.gz' 'val-metadata.tar.gz'; do
+    wget "${repository_url}/resolve/main/${archive}"
+done
+
+echo "Extracting archives..."
+for archive in 'train-images.tar.gz' 'train-metadata.tar.gz' 'val-images.tar.gz' 'val-metadata.tar.gz'; do
+  if command -v pv &> /dev/null; then
+    pv "${archive}" | tar -xz
+  else
+    tar -xz -f "${archive}"
+  fi
+done
+
+echo "Downloading split files..."
+for split in 'train' 'val'; do
+    wget "${repository_url}/resolve/main/splits/${split}.txt"
+done
+
+cd - &> /dev/null