updates

scripts/make_panel_sweep_contact_sheets.py

@@ -0,0 +1,135 @@
+#!/usr/bin/env python3
+"""Build labeled contact sheets from serial_scenario webcam snapshots."""
+
+from __future__ import annotations
+
+import argparse
+import math
+from pathlib import Path
+
+import cv2
+import numpy as np
+
+
+CROP_PRESETS = {
+    "full": (0.00, 0.00, 1.00, 1.00),
+    "panel": (0.03, 0.10, 0.98, 0.94),
+    "right-stack": (0.70, 0.23, 0.92, 0.86),
+    "lcd": (0.00, 0.22, 0.38, 0.78),
+}
+
+
+def main() -> int:
+    args = build_arg_parser().parse_args()
+    images = sorted(args.snapshot_dir.glob(args.glob))
+    if args.only_candidates:
+        images = [path for path in images if "candidate_" in path.name]
+    if not images:
+        raise SystemExit(f"no images found in {args.snapshot_dir}")
+
+    output_dir = args.output_dir or args.snapshot_dir.with_name(args.snapshot_dir.name + "-sheets")
+    output_dir.mkdir(parents=True, exist_ok=True)
+
+    per_sheet = args.cols * args.rows
+    total_sheets = math.ceil(len(images) / per_sheet)
+    written: list[Path] = []
+    for sheet_index in range(total_sheets):
+        group = images[sheet_index * per_sheet : (sheet_index + 1) * per_sheet]
+        sheet = build_sheet(group, args)
+        out = output_dir / f"sheet-{sheet_index + 1:03d}.jpg"
+        cv2.imwrite(str(out), sheet)
+        written.append(out)
+
+    print(f"images={len(images)} sheets={len(written)} output_dir={output_dir}")
+    if written:
+        print(f"first={written[0]}")
+        print(f"last={written[-1]}")
+    return 0
+
+
+def build_arg_parser() -> argparse.ArgumentParser:
+    parser = argparse.ArgumentParser(description="Create labeled contact sheets from webcam snapshots.")
+    parser.add_argument("snapshot_dir", type=Path, help="directory containing serial_scenario snapshots")
+    parser.add_argument("--output-dir", type=Path, help="directory for generated contact sheets")
+    parser.add_argument("--glob", default="*.jpg", help="image glob within snapshot_dir")
+    parser.add_argument("--cols", type=int, default=4, help="thumbnail columns per sheet")
+    parser.add_argument("--rows", type=int, default=5, help="thumbnail rows per sheet")
+    parser.add_argument("--thumb-width", type=int, default=360, help="thumbnail image width")
+    parser.add_argument("--label-height", type=int, default=48, help="label area height per thumbnail")
+    parser.add_argument(
+        "--crop",
+        choices=sorted(CROP_PRESETS),
+        default="panel",
+        help="crop preset before thumbnailing",
+    )
+    parser.add_argument(
+        "--only-candidates",
+        action="store_true",
+        help="include only filenames containing candidate_",
+    )
+    return parser
+
+
+def build_sheet(paths: list[Path], args: argparse.Namespace) -> np.ndarray:
+    cells = [build_cell(path, args) for path in paths]
+    blank = np.full_like(cells[0], 245)
+    while len(cells) < args.cols * args.rows:
+        cells.append(blank.copy())
+
+    rows = []
+    for row_index in range(args.rows):
+        row_cells = cells[row_index * args.cols : (row_index + 1) * args.cols]
+        rows.append(np.hstack(row_cells))
+    return np.vstack(rows)
+
+
+def build_cell(path: Path, args: argparse.Namespace) -> np.ndarray:
+    image = cv2.imread(str(path))
+    if image is None:
+        image = np.full((240, 320, 3), 220, dtype=np.uint8)
+    image = crop_image(image, CROP_PRESETS[args.crop])
+    thumb_height = max(1, int(image.shape[0] * args.thumb_width / max(1, image.shape[1])))
+    thumb = cv2.resize(image, (args.thumb_width, thumb_height), interpolation=cv2.INTER_AREA)
+
+    label = snapshot_label(path)
+    cell = np.full((args.label_height + thumb_height, args.thumb_width, 3), 245, dtype=np.uint8)
+    cell[args.label_height : args.label_height + thumb_height, 0 : args.thumb_width] = thumb
+    draw_label(cell, label, args.label_height)
+    return cell
+
+
+def crop_image(image: np.ndarray, crop: tuple[float, float, float, float]) -> np.ndarray:
+    h, w = image.shape[:2]
+    x1, y1, x2, y2 = crop
+    left = min(w - 1, max(0, int(w * x1)))
+    top = min(h - 1, max(0, int(h * y1)))
+    right = min(w, max(left + 1, int(w * x2)))
+    bottom = min(h, max(top + 1, int(h * y2)))
+    return image[top:bottom, left:right]
+
+
+def draw_label(cell: np.ndarray, text: str, label_height: int) -> None:
+    font = cv2.FONT_HERSHEY_SIMPLEX
+    scale = 0.48
+    thickness = 1
+    max_chars = 45
+    line1 = text[:max_chars]
+    line2 = text[max_chars : max_chars * 2]
+    cv2.putText(cell, line1, (6, 18), font, scale, (0, 0, 0), thickness, cv2.LINE_AA)
+    if line2:
+        cv2.putText(cell, line2, (6, min(label_height - 8, 38)), font, scale, (0, 0, 0), thickness, cv2.LINE_AA)
+
+
+def snapshot_label(path: Path) -> str:
+    name = path.name
+    if "_tx_0500ms_" in name:
+        label = name.split("_tx_0500ms_", 1)[1]
+    elif "_tx_" in name:
+        label = name.split("_tx_", 1)[1]
+    else:
+        label = name
+    return label.rsplit("_", 1)[0]
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())