328 lines
12 KiB
Python
328 lines
12 KiB
Python
import glob
|
|
import os
|
|
import random
|
|
import cv2
|
|
|
|
# --- CONFIGURATION ---
|
|
OUTPUT_DIRECTORY = "dataset"
|
|
FRAMES_TO_EXTRACT = 20 # Number of random anchor frames to find targets
|
|
SCALE_FACTOR = 0.5 # Resize video to 1/2 size for easier viewing
|
|
|
|
# Define the EXACT size of the bounding box around your click point
|
|
BOX_WIDTH = 64
|
|
BOX_HEIGHT = 64
|
|
|
|
# Defined classification options (YOLO maps these to IDs: 0, 1, 2, 3...)
|
|
CLASSIFICATION_OPTIONS = [
|
|
"guardian",
|
|
"interceptor",
|
|
"sentinel",
|
|
"rocket",
|
|
"bomber",
|
|
"background",
|
|
"asteroid",
|
|
]
|
|
# ---------------------
|
|
|
|
def get_dataset_counts():
|
|
"""Counts the number of existing bounding boxes by parsing YOLO text files."""
|
|
counts = {option: 0 for option in CLASSIFICATION_OPTIONS}
|
|
labels_dir = os.path.join(OUTPUT_DIRECTORY, "labels")
|
|
|
|
if os.path.exists(labels_dir):
|
|
# Scan all text annotation files in the YOLO labels directory
|
|
for txt_file in glob.glob(os.path.join(labels_dir, "*.txt")):
|
|
try:
|
|
with open(txt_file, "r") as f:
|
|
for line in f:
|
|
parts = line.strip().split()
|
|
if parts:
|
|
class_id = int(parts[0])
|
|
if 0 <= class_id < len(CLASSIFICATION_OPTIONS):
|
|
option = CLASSIFICATION_OPTIONS[class_id]
|
|
counts[option] += 1
|
|
except Exception:
|
|
pass # Skip corrupted or unreadable text files safely
|
|
return counts
|
|
|
|
def mouse_click_callback(event, x, y, flags, param):
|
|
"""Handles mouse clicks.
|
|
|
|
Calculates YOLO coordinates, saves the full frame, and appends a line to
|
|
the text file.
|
|
"""
|
|
if event == cv2.EVENT_LBUTTONDOWN:
|
|
# Ignore clicks if the user is in "background" capture mode
|
|
if param["session_label"] == "background":
|
|
print(
|
|
" Info: In background mode, do not click targets. Just press SPACE to save frame."
|
|
)
|
|
return
|
|
|
|
frame = param["frame"]
|
|
display_frame = param["display_frame"]
|
|
frame_idx = param["frame_idx"]
|
|
window_name = param["window_name"]
|
|
|
|
# Calculate top-left and bottom-right corners
|
|
x1 = x - BOX_WIDTH // 2
|
|
y1 = y - BOX_HEIGHT // 2
|
|
x2 = x1 + BOX_WIDTH
|
|
y2 = y1 + BOX_HEIGHT
|
|
|
|
# --- Edge Case Handling ---
|
|
if x1 < 0:
|
|
x1 = 0
|
|
x2 = BOX_WIDTH
|
|
if y1 < 0:
|
|
y1 = 0
|
|
y2 = BOX_HEIGHT
|
|
if x2 > frame.shape[1]:
|
|
x2 = frame.shape[1]
|
|
x1 = x2 - BOX_WIDTH
|
|
if y2 > frame.shape[0]:
|
|
y2 = frame.shape[0]
|
|
y1 = y2 - BOX_HEIGHT
|
|
|
|
# --- YOLO Normalized Coordinate Conversion ---
|
|
img_h, img_w = frame.shape[0], frame.shape[1]
|
|
|
|
true_center_x = (x1 + x2) / 2.0
|
|
true_center_y = (y1 + y2) / 2.0
|
|
|
|
x_center_norm = true_center_x / img_w
|
|
y_center_norm = true_center_y / img_h
|
|
width_norm = BOX_WIDTH / img_w
|
|
height_norm = BOX_HEIGHT / img_h
|
|
|
|
# Clean video name to ensure a unique, safe filename
|
|
video_base = os.path.splitext(os.path.basename(param["video_name"]))[0]
|
|
filename_base = f"{video_base}_frame_{frame_idx}"
|
|
|
|
image_save_path = os.path.join(
|
|
param["images_dir"], f"{filename_base}.jpg"
|
|
)
|
|
label_save_path = os.path.join(
|
|
param["labels_dir"], f"{filename_base}.txt"
|
|
)
|
|
|
|
# 1. Save the full frame image (overwriting is fine if clicking multiple items)
|
|
cv2.imwrite(image_save_path, frame)
|
|
|
|
# 2. Append the target annotation to the YOLO text file
|
|
class_id = param["class_id"]
|
|
with open(label_save_path, "a") as f:
|
|
f.write(
|
|
f"{class_id} {x_center_norm:.6f} {y_center_norm:.6f} {width_norm:.6f} {height_norm:.6f}\n"
|
|
)
|
|
|
|
print(
|
|
f" Logged {param['session_label']} box coordinate to {label_save_path}"
|
|
)
|
|
param["box_drawn"] = True
|
|
|
|
# Draw visual feedback box and short class label on screen
|
|
cv2.rectangle(display_frame, (x1, y1), (x2, y2), (0, 255, 0), 2)
|
|
cv2.putText(
|
|
display_frame,
|
|
param["session_label"],
|
|
(x1, y1 - 5),
|
|
cv2.FONT_HERSHEY_SIMPLEX,
|
|
0.4,
|
|
(0, 255, 0),
|
|
1,
|
|
)
|
|
cv2.imshow(window_name, display_frame)
|
|
|
|
def label_video_session():
|
|
print("=========================================")
|
|
print(" AI YOLO DATASET LABELER ")
|
|
print("=========================================\n")
|
|
|
|
# 1. Automatically find and select a random MP4 video
|
|
mp4_files = glob.glob("*.mp4")
|
|
if not mp4_files:
|
|
print(
|
|
"Error: No .mp4 files found in the current directory.\n"
|
|
"Please place this script in the same folder as your video files."
|
|
)
|
|
return
|
|
|
|
selected_video = random.choice(mp4_files)
|
|
print(f"-> Selected Video File: '{selected_video}'")
|
|
|
|
# 2. Fetch and display current dataset balance dashboard
|
|
counts = get_dataset_counts()
|
|
print("\n-----------------------------------------")
|
|
print(" CURRENT DATASET BALANCE STATUS ")
|
|
print("-----------------------------------------")
|
|
for option, count in counts.items():
|
|
print(f" • {option.capitalize():<12} : {count} boxes logged")
|
|
print("-----------------------------------------")
|
|
print(
|
|
"💡 Tip: Try to choose options with lower counts to keep data balanced!"
|
|
)
|
|
|
|
# 3. Force selection of a valid label
|
|
print("\nAvailable Classification Targets:")
|
|
for idx, option in enumerate(CLASSIFICATION_OPTIONS, start=1):
|
|
print(f" [{idx}] {option} (Current Boxes: {counts[option]})")
|
|
|
|
while True:
|
|
try:
|
|
choice = input(
|
|
f"\nEnter the number (1-{len(CLASSIFICATION_OPTIONS)}) of what you are labeling: "
|
|
).strip()
|
|
choice_idx = int(choice) - 1
|
|
if 0 <= choice_idx < len(CLASSIFICATION_OPTIONS):
|
|
session_label = CLASSIFICATION_OPTIONS[choice_idx]
|
|
session_class_id = choice_idx
|
|
break
|
|
else:
|
|
print("Invalid selection. Please choose a number from the list.")
|
|
except ValueError:
|
|
print("Invalid input. Please enter a number.")
|
|
|
|
# Open video and check validity
|
|
cap = cv2.VideoCapture(selected_video)
|
|
if not cap.isOpened():
|
|
print(f"Error: Could not open video file {selected_video}")
|
|
return
|
|
|
|
total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
|
|
if total_frames == 0:
|
|
print("Error: Video has 0 frames.")
|
|
return
|
|
|
|
# Select random unique anchor frames to jump between
|
|
num_frames = min(FRAMES_TO_EXTRACT, total_frames)
|
|
frame_indices = random.sample(range(total_frames), num_frames)
|
|
frame_indices.sort()
|
|
|
|
# Set up standard YOLO dataset directory branches
|
|
images_dir = os.path.join(OUTPUT_DIRECTORY, "images")
|
|
labels_dir = os.path.join(OUTPUT_DIRECTORY, "labels")
|
|
os.makedirs(images_dir, exist_ok=True)
|
|
os.makedirs(labels_dir, exist_ok=True)
|
|
|
|
# Comprehensive User Instructions
|
|
print("\n" + "=" * 50)
|
|
print(" HOW TO LABEL")
|
|
print("=" * 50)
|
|
print(f"• TARGET OBJECT: {session_label.upper()}")
|
|
print(f"• SESSION SCOPE: {num_frames} random search zones in '{selected_video}'")
|
|
print(f"• BOUNDING BOX : {BOX_WIDTH}x{BOX_HEIGHT} pixels scaled to YOLO layout")
|
|
print("-" * 50)
|
|
print("► CONTROLS:")
|
|
print(" [Left-Click] - Click directly on target to tag position (auto-saves frame).")
|
|
print(" [Right Arrow] - Move FORWARD 1 individual frame (sequential tracking, no save).")
|
|
print(" [Left Arrow] - Move BACKWARD 1 individual frame (sequential tracking, no save).")
|
|
print(" [SPACE / ENTER] - Jump completely to the NEXT random video search spot.")
|
|
print(" [Q Key] - Quit and save all progress up to this point.")
|
|
print("=" * 50)
|
|
input("\nPress ENTER when you are ready to begin...")
|
|
|
|
# Cross-platform arrow key mappings for waitKeyEx()
|
|
LEFT_KEYS = [81, 2, 2424832, 65361, 63234]
|
|
RIGHT_KEYS = [83, 3, 2490368, 65363, 63235]
|
|
|
|
i = 0
|
|
current_frame = frame_indices[i]
|
|
|
|
while i < len(frame_indices):
|
|
cap.set(cv2.CAP_PROP_POS_FRAMES, current_frame)
|
|
ret, frame = cap.read()
|
|
if not ret:
|
|
print(f" Warning: Could not read frame {current_frame}. Moving to next random anchor.")
|
|
i += 1
|
|
if i < len(frame_indices):
|
|
current_frame = frame_indices[i]
|
|
continue
|
|
|
|
# Resize the frame by 1/2 for laptop viewability
|
|
frame_resized = cv2.resize(
|
|
frame, (0, 0), fx=SCALE_FACTOR, fy=SCALE_FACTOR
|
|
)
|
|
frame_display = frame_resized.copy()
|
|
|
|
window_name = f"Labeling: {session_label} | Spot {i+1}/{num_frames} (Frame {current_frame})"
|
|
cv2.namedWindow(window_name)
|
|
|
|
session_state = {
|
|
"frame": frame_resized,
|
|
"display_frame": frame_display,
|
|
"images_dir": images_dir,
|
|
"labels_dir": labels_dir,
|
|
"frame_idx": current_frame,
|
|
"window_name": window_name,
|
|
"video_name": selected_video,
|
|
"class_id": session_class_id,
|
|
"session_label": session_label,
|
|
"box_drawn": False,
|
|
}
|
|
|
|
cv2.setMouseCallback(window_name, mouse_click_callback, session_state)
|
|
|
|
# Keep window open until user acts
|
|
cv2.imshow(window_name, frame_display)
|
|
while True:
|
|
cv2.moveWindow(window_name, 0, 0)
|
|
key = cv2.waitKeyEx(1)
|
|
|
|
# Enter (13) or Space (32) -> Jump to the next random anchor spot
|
|
if key == 13 or key == 32:
|
|
if session_label == "background":
|
|
video_base = os.path.splitext(
|
|
os.path.basename(selected_video)
|
|
)[0]
|
|
filename_base = f"{video_base}_frame_{current_frame}"
|
|
image_save_path = os.path.join(
|
|
images_dir, f"{filename_base}.jpg"
|
|
)
|
|
label_save_path = os.path.join(
|
|
labels_dir, f"{filename_base}.txt"
|
|
)
|
|
|
|
cv2.imwrite(image_save_path, frame_resized)
|
|
open(label_save_path, "a").close() # Creates empty file
|
|
print(f" Registered negative background frame: {filename_base}")
|
|
|
|
i += 1
|
|
if i < len(frame_indices):
|
|
current_frame = frame_indices[i]
|
|
break
|
|
|
|
# Right Arrow -> Move forward exactly ONE individual sequential frame
|
|
elif key in RIGHT_KEYS:
|
|
if current_frame < total_frames - 10:
|
|
current_frame += 10
|
|
else:
|
|
print(" Already at the very last frame of the video!")
|
|
continue
|
|
break
|
|
|
|
# Left Arrow -> Move backward exactly ONE individual sequential frame
|
|
elif key in LEFT_KEYS:
|
|
if current_frame > 0:
|
|
current_frame -= 10
|
|
else:
|
|
print(" Already at the first frame of the video!")
|
|
continue
|
|
break
|
|
|
|
# Quit
|
|
elif key == ord("q") or key == ord("Q") or (key & 0xFF) in [ord("q"), ord("Q")]:
|
|
print("\nQuitting session early... Saving progress.")
|
|
cap.release()
|
|
cv2.destroyAllWindows()
|
|
return
|
|
|
|
cv2.destroyWindow(window_name)
|
|
|
|
cap.release()
|
|
cv2.destroyAllWindows()
|
|
print(f"\nSession finished! Dataset updated successfully inside '{OUTPUT_DIRECTORY}'.")
|
|
|
|
if __name__ == "__main__":
|
|
label_video_session()
|