File size: 12,337 Bytes
db613b6 1150050 db613b6 1150050 db613b6 1150050 db613b6 1150050 db613b6 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 | """Command-line entry point: `kitti-c2l generate|verify|stats`."""
from __future__ import annotations
import argparse
import sys
from collections.abc import Iterator
from pathlib import Path
import pyarrow.parquet as pq
from kitti_c2l.assets import AssetRow, generate_sequence_assets
from kitti_c2l.generate import DEFAULT_BASE_SEED, DEFAULT_CAMERA_ID, generate_sequence_rows
from kitti_c2l.sequence_mapping import AVAILABLE_SEQUENCES, SequenceInfo, load_sequence_mapping
from kitti_c2l.stages import STAGES
from kitti_c2l.windows import (
DEFAULT_WINDOW_LENGTH,
DEFAULT_WINDOWS_BASE_SEED,
generate_sequence_windows,
)
from kitti_c2l.writer import write_assets_sharded, write_split, write_windows
def _cmd_generate(args: argparse.Namespace) -> int:
seq_map = load_sequence_mapping()
user_specified = args.sequences is not None
sequences = args.sequences.split(",") if user_specified else list(AVAILABLE_SEQUENCES)
by_split: dict[str, list] = {}
for seq_id in sequences:
if seq_id not in seq_map:
print(f"Unknown sequence {seq_id!r}, skipping.", file=sys.stderr)
continue
seq_info = seq_map[seq_id]
if not seq_info.available and not user_specified:
print(
f"sequence {seq_id}: SKIPPED -- raw drive not publicly available "
f"(see configs/sequence_mapping.yaml). Pass --sequences {seq_id} explicitly to force.",
file=sys.stderr,
)
continue
frame_indices = list(range(args.max_frames)) if args.max_frames else None
rows = list(
generate_sequence_rows(
seq_info,
args.kitti_calib_root,
frame_indices=frame_indices,
stages=STAGES,
camera_id=args.camera_id,
base_seed=args.base_seed,
camera_model=args.camera_model,
)
)
by_split.setdefault(seq_info.split, []).extend(rows)
print(f"sequence {seq_id}: generated {len(rows)} rows ({seq_info.split})")
total = 0
for split, rows in by_split.items():
output_path = Path(args.output_dir) / f"{split}.parquet"
count = write_split(rows, output_path)
total += count
print(f"wrote {count} rows -> {output_path}")
print(f"total: {total} rows")
return 0
def _cmd_generate_assets(args: argparse.Namespace) -> int:
seq_map = load_sequence_mapping()
user_specified = args.sequences is not None
sequences = args.sequences.split(",") if user_specified else list(AVAILABLE_SEQUENCES)
by_split: dict[str, list] = {}
for seq_id in sequences:
if seq_id not in seq_map:
print(f"Unknown sequence {seq_id!r}, skipping.", file=sys.stderr)
continue
seq_info = seq_map[seq_id]
if not seq_info.available and not user_specified:
print(
f"sequence {seq_id}: SKIPPED -- raw drive not publicly available "
f"(see configs/sequence_mapping.yaml). Pass --sequences {seq_id} explicitly to force.",
file=sys.stderr,
)
continue
by_split.setdefault(seq_info.split, []).append(seq_info)
def _rows_for_split(seq_infos: list[SequenceInfo]) -> Iterator[AssetRow]:
frame_indices = list(range(args.max_frames)) if args.max_frames else None
for seq_info in seq_infos:
count = 0
for row in generate_sequence_assets(
seq_info, args.kitti_raw_root, frame_indices=frame_indices, camera_id=args.camera_id
):
count += 1
yield row
print(
f"sequence {seq_info.sequence}: materialized {count} frame assets ({seq_info.split})"
)
total = 0
for split, seq_infos in by_split.items():
shard_paths = write_assets_sharded(_rows_for_split(seq_infos), args.output_dir, split)
total_split_rows = sum(pq.read_metadata(p).num_rows for p in shard_paths)
total += total_split_rows
print(f"wrote {total_split_rows} rows across {len(shard_paths)} shard(s) for split={split}")
print(f"total: {total} frame assets")
return 0
def _cmd_generate_windows(args: argparse.Namespace) -> int:
seq_map = load_sequence_mapping()
user_specified = args.sequences is not None
sequences = args.sequences.split(",") if user_specified else list(AVAILABLE_SEQUENCES)
by_split: dict[str, list] = {}
for seq_id in sequences:
if seq_id not in seq_map:
print(f"Unknown sequence {seq_id!r}, skipping.", file=sys.stderr)
continue
seq_info = seq_map[seq_id]
if not seq_info.available and not user_specified:
print(
f"sequence {seq_id}: SKIPPED -- raw drive not publicly available "
f"(see configs/sequence_mapping.yaml). Pass --sequences {seq_id} explicitly to force.",
file=sys.stderr,
)
continue
rows = list(
generate_sequence_windows(
seq_info,
args.kitti_calib_root,
window_length=args.window_length,
stages=STAGES,
camera_id=args.camera_id,
base_seed=args.base_seed,
max_windows=args.max_windows,
camera_model=args.camera_model,
)
)
num_windows = len(rows) // len(STAGES)
by_split.setdefault(seq_info.split, []).extend(rows)
print(
f"sequence {seq_id}: generated {num_windows} windows ({len(rows)} rows, {seq_info.split})"
)
total = 0
for split, rows in by_split.items():
output_path = Path(args.output_dir) / f"windows_{split}.parquet"
count = write_windows(rows, output_path)
total += count
print(f"wrote {count} rows -> {output_path}")
print(f"total: {total} rows")
return 0
def _cmd_stats(args: argparse.Namespace) -> int:
seq_map = load_sequence_mapping()
total_frames = 0
available_frames = 0
for seq_id, seq_info in seq_map.items():
tag = "" if seq_info.available else " [NOT PUBLICLY AVAILABLE -- excluded by default]"
print(
f"{seq_id}: {seq_info.date}/drive_{seq_info.drive} split={seq_info.split} "
f"frames={seq_info.frame_count}{tag}"
)
total_frames += seq_info.frame_count
if seq_info.available:
available_frames += seq_info.frame_count
print(
f"total frames (11 sequences): {total_frames}, total rows at {len(STAGES)} stages: {total_frames * len(STAGES)}"
)
print(
f"available frames (10 sequences, default `generate`): {available_frames}, "
f"total rows at {len(STAGES)} stages: {available_frames * len(STAGES)}"
)
return 0
def _cmd_validate(args: argparse.Namespace) -> int:
import json
from kitti_c2l.release import validate_release
report = validate_release(
args.root, kitti_calib_root=args.kitti_calib_root, check_manifest=not args.write_manifest
)
if args.write_manifest:
destination = args.root / "release/v1/manifest.json"
destination.parent.mkdir(parents=True, exist_ok=True)
destination.write_text(json.dumps(report, indent=2) + "\n")
print(json.dumps(report, indent=2))
return 0
def build_parser() -> argparse.ArgumentParser:
parser = argparse.ArgumentParser(prog="kitti-c2l")
subparsers = parser.add_subparsers(dest="command", required=True)
generate_parser = subparsers.add_parser(
"generate", help="Generate parquet metadata for one or more sequences."
)
generate_parser.add_argument(
"--kitti-calib-root",
required=True,
help="Directory containing per-date KITTI raw calibration subdirectories (e.g. .../2011_09_30/calib_*.txt).",
)
generate_parser.add_argument(
"--output-dir", default="data", help="Directory to write {split}.parquet into."
)
generate_parser.add_argument(
"--sequences",
default=None,
help="Comma-separated sequence IDs, default: all publicly-available ones (excludes 03; pass "
"--sequences 03 explicitly to force an attempt).",
)
generate_parser.add_argument("--camera-id", type=int, default=DEFAULT_CAMERA_ID)
generate_parser.add_argument(
"--camera-model",
choices=("legacy_v1", "full_projection_v1_1"),
default="legacy_v1",
help="Calibration conversion: frozen v1 x-only or source-consistent full projection (v1.1).",
)
generate_parser.add_argument("--base-seed", type=int, default=DEFAULT_BASE_SEED)
generate_parser.add_argument(
"--max-frames",
type=int,
default=None,
help="Cap frames per sequence (for quick test slices).",
)
generate_parser.set_defaults(func=_cmd_generate)
assets_parser = subparsers.add_parser(
"generate-assets",
help="Materialize the fully self-contained release: image + raw LiDAR bytes, once per unique frame.",
)
assets_parser.add_argument(
"--kitti-raw-root",
required=True,
help="Directory containing per-date KITTI raw subdirectories (calib files + <date>_drive_<drive>_sync/).",
)
assets_parser.add_argument(
"--output-dir", default="data", help="Directory to write frames_{split}.parquet into."
)
assets_parser.add_argument(
"--sequences", default=None, help="Same semantics as `generate --sequences`."
)
assets_parser.add_argument("--camera-id", type=int, default=DEFAULT_CAMERA_ID)
assets_parser.add_argument(
"--max-frames",
type=int,
default=None,
help="Cap frames per sequence (for quick test slices).",
)
assets_parser.set_defaults(func=_cmd_generate_assets)
windows_parser = subparsers.add_parser(
"generate-windows",
help="Generate sequence-window metadata: one decalibration held constant across each window's frames.",
)
windows_parser.add_argument(
"--kitti-calib-root",
required=True,
help="Directory containing per-date KITTI raw calibration subdirectories (e.g. .../2011_09_30/calib_*.txt).",
)
windows_parser.add_argument(
"--output-dir", default="data", help="Directory to write windows_{split}.parquet into."
)
windows_parser.add_argument(
"--sequences", default=None, help="Same semantics as `generate --sequences`."
)
windows_parser.add_argument("--camera-id", type=int, default=DEFAULT_CAMERA_ID)
windows_parser.add_argument(
"--camera-model",
choices=("legacy_v1", "full_projection_v1_1"),
default="legacy_v1",
help="Calibration conversion; use the same model for frame and window metadata.",
)
windows_parser.add_argument("--base-seed", type=int, default=DEFAULT_WINDOWS_BASE_SEED)
windows_parser.add_argument(
"--window-length",
type=int,
default=DEFAULT_WINDOW_LENGTH,
help=f"Frames per non-overlapping window (default: {DEFAULT_WINDOW_LENGTH}, ~3s at KITTI's 10Hz).",
)
windows_parser.add_argument(
"--max-windows",
type=int,
default=None,
help="Cap windows per sequence (for quick test slices).",
)
windows_parser.set_defaults(func=_cmd_generate_windows)
validate_parser = subparsers.add_parser(
"validate", help="Audit the frozen metadata release and its checksums."
)
validate_parser.add_argument("--root", type=Path, default=Path("."))
validate_parser.add_argument("--kitti-calib-root", type=Path)
validate_parser.add_argument(
"--write-manifest",
action="store_true",
help="Write a manifest after a complete successful audit.",
)
validate_parser.set_defaults(func=_cmd_validate)
stats_parser = subparsers.add_parser(
"stats", help="Print sequence mapping stats without touching any KITTI data."
)
stats_parser.set_defaults(func=_cmd_stats)
return parser
def main(argv: list[str] | None = None) -> int:
parser = build_parser()
args = parser.parse_args(argv)
return args.func(args)
if __name__ == "__main__":
sys.exit(main())
|