File size: 4,031 Bytes
c2b4e9c | 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 | """
Convert YOLO-segmentation-format labels (polygons) to YOLO-detection-format
labels (bounding boxes), so a segmentation-labeled dataset can be dropped
into a standard detection training pipeline.
YOLO segmentation label line format:
class_id x1 y1 x2 y2 x3 y3 ... xn yn (all normalized 0-1)
YOLO detection label line format:
class_id x_center y_center width height (all normalized 0-1)
USAGE:
Just set SRC_LABEL_DIR below to the folder containing the segmentation
.txt files, and DST_LABEL_DIR to where you want the converted detection
.txt files written. The script handles the rest -- multiple objects per
file, multiple classes, etc.
If you want to convert in place (overwrite the same files), set
DST_LABEL_DIR equal to SRC_LABEL_DIR -- but a backup is strongly
recommended first since this is a destructive format change.
"""
import os
import shutil
# ------------------------- CONFIG (edit these) -------------------------
SRC_LABEL_DIR = r"C:\Users\shrey\Downloads\project1.yolo26 (1)\valid\labels" # <-- drop your segmentation labels folder path here
DST_LABEL_DIR = r"C:\Users\shrey\Downloads\dataset\labels_1" # <-- converted bbox labels will be written here
# If True, skips any line that doesn't look like a valid polygon (odd number
# of coordinates, too few points, etc.) instead of crashing the whole run.
SKIP_INVALID_LINES = True
# -------------------------------------------------------------------------
def polygon_to_bbox(coords):
"""coords: flat list [x1, y1, x2, y2, ...] normalized 0-1. Returns (xc, yc, w, h)."""
xs = coords[0::2]
ys = coords[1::2]
x_min, x_max = min(xs), max(xs)
y_min, y_max = min(ys), max(ys)
x_center = (x_min + x_max) / 2
y_center = (y_min + y_max) / 2
width = x_max - x_min
height = y_max - y_min
return x_center, y_center, width, height
def convert_file(src_path, dst_path):
converted_lines = []
skipped = 0
with open(src_path) as f:
for line in f:
line = line.strip()
if not line:
continue
parts = line.split()
class_id = parts[0]
coords = list(map(float, parts[1:]))
if len(coords) < 6 or len(coords) % 2 != 0:
# need at least 3 points (6 values) for a valid polygon
skipped += 1
if not SKIP_INVALID_LINES:
raise ValueError(f"Invalid polygon in {src_path}: {line}")
continue
xc, yc, w, h = polygon_to_bbox(coords)
converted_lines.append(f"{class_id} {xc:.6f} {yc:.6f} {w:.6f} {h:.6f}")
with open(dst_path, "w") as f:
f.write("\n".join(converted_lines) + ("\n" if converted_lines else ""))
return len(converted_lines), skipped
def main():
if not os.path.isdir(SRC_LABEL_DIR):
print(f"Source folder not found: {SRC_LABEL_DIR}")
return
os.makedirs(DST_LABEL_DIR, exist_ok=True)
total_files = 0
total_boxes = 0
total_skipped = 0
for fname in os.listdir(SRC_LABEL_DIR):
if not fname.endswith(".txt"):
continue
src_path = os.path.join(SRC_LABEL_DIR, fname)
dst_path = os.path.join(DST_LABEL_DIR, fname)
n_boxes, n_skipped = convert_file(src_path, dst_path)
total_files += 1
total_boxes += n_boxes
total_skipped += n_skipped
print(f"Converted {total_files} label file(s).")
print(f"Total bounding boxes written: {total_boxes}")
if total_skipped:
print(f"Skipped {total_skipped} invalid/malformed polygon line(s) -- check these manually if the count seems high.")
print(f"\nOutput written to: {DST_LABEL_DIR}")
print("Next: point your data.yaml / dataset folder at these converted labels "
"(alongside the original images) before training.")
if __name__ == "__main__":
main() |