chore(docs): document ScoreExtractor tiling and refactor debug scripts (#563)

scripts/debug/test_live_ocr.py

@@ -0,0 +1,75 @@
+import cv2
+import numpy as np
+import easyocr
+import re
+from youtube_tab_to_pdf import TemporalTracker
+
+cap = cv2.VideoCapture(r"C:\Users\Certes\Desktop\guitar_score\output\shintakarajima.mp4")
+total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
+fps = cap.get(cv2.CAP_PROP_FPS)
+
+tracker = TemporalTracker()
+cap.set(cv2.CAP_PROP_POS_FRAMES, 500)
+ret, check_frame = cap.read()
+
+from youtube_tab_to_pdf import _find_white_tab_strip
+bounds = _find_white_tab_strip(cv2.resize(check_frame, (1280, int(check_frame.shape[0] * (1280/check_frame.shape[1])))))
+if bounds:
+    crop_top = max(0, bounds[0] - 60)
+    crop_bottom = bounds[1]
+    tracker.set_crop(crop_top, crop_bottom)
+
+# Process only first 95 seconds to get unique pages
+print("Extracting unique pages from first 95 seconds...")
+tracker.process_video(cap, start_sec=0, end_sec=95)
+unique_pages = tracker.get_unique_pages()
+
+print(f"Extracted {len(unique_pages)} unique pages.")
+
+# Try easyOCR
+reader = easyocr.Reader(['en'], gpu=False)
+
+def extract_measure_number(page_bgr):
+    cw = min(page_bgr.shape[1], 1000)
+    page_gray = cv2.cvtColor(page_bgr[:, :cw], cv2.COLOR_BGR2GRAY)
+    _, bin_inv = cv2.threshold(page_gray, 200, 255, cv2.THRESH_BINARY_INV)
+    row_sums = np.sum(bin_inv, axis=1) / 255.0
+    staff_rows = np.where(row_sums > cw * 0.4)[0]
+    
+    if len(staff_rows) >= 6:
+        staff_y_top, staff_y_bottom = staff_rows[0], staff_rows[-1]
+        for r in staff_rows:
+            if r - staff_y_top > 100: break
+            staff_y_bottom = r
+    else:
+        return -1
+        
+    expected_h = max(10, staff_y_bottom - staff_y_top + 1)
+    staff_region = bin_inv[staff_y_top:staff_y_bottom+1, :]
+    col_sums = np.sum(staff_region, axis=0) / 255.0
+    bar_xs = np.where(col_sums >= expected_h * 0.8)[0]
+    
+    if len(bar_xs) == 0: return -1
+    x_bar = bar_xs[0]
+    
+    box_y1 = max(0, staff_y_top - 25)
+    box_y2 = staff_y_top
+    box_x1 = x_bar
+    box_x2 = min(page_gray.shape[1], x_bar + 35)
+    
+    num_box = page_gray[box_y1:box_y2, box_x1:box_x2]
+    _, num_inv = cv2.threshold(num_box, 200, 255, cv2.THRESH_BINARY_INV)
+    num_for_ocr = cv2.bitwise_not(num_inv) 
+    
+    upscaled = cv2.resize(num_for_ocr, None, fx=4, fy=4, interpolation=cv2.INTER_CUBIC)
+    padded = cv2.copyMakeBorder(upscaled, 20, 20, 20, 20, cv2.BORDER_CONSTANT, value=[255, 255, 255])
+    
+    results = reader.readtext(padded, allowlist="0123456789")
+    if not results: return -1
+    
+    digits = re.findall(r'\d+', results[0][1])
+    return int(digits[0]) if digits else -1
+
+for i, page in enumerate(unique_pages):
+    num = extract_measure_number(page)
+    print(f"Page {i:02d}: {num}")