Extract validation set from IMAGENET into another zip file, with label-directories

imagenet-extract-val.py

import os.path
import gzip
import pickle
from zipfile import ZipFile

# from https://github.com/onnx/models/blob/8d50e3f/vision/classification/imagenet_val_maps.pklz
with gzip.open('imagenet_val_maps.pklz', 'rb') as f:
    dirs, mappings = pickle.load(f)
src_dir = "ILSVRC/Data/CLS-LOC/val"
target_dir = "ILSVRC/val"
path_map = {}
for m in mappings:
    path_map[os.path.join(src_dir, m[0])] = os.path.join(target_dir, m[1], m[0])

# from https://www.kaggle.com/competitions/imagenet-object-localization-challenge/data
with ZipFile('imagenet-val.zip', 'w') as ozf:
    with ZipFile('imagenet.zip', 'r') as zf:
        dirs = set()
        for path in zf.namelist():
            if path.startswith("ILSVRC/Data/CLS-LOC/val/"):
                newpath = path_map[path]
                ozf.writestr(newpath, zf.read(path))
            elif path == "LOC_synset_mapping.txt":
                ozf.writestr(f"ILSVRC/synset_mapping.txt", zf.read(path))
            elif path == "LOC_val_solution.csv":
                ozf.writestr(f"ILSVRC/val_loc.csv", zf.read(path))