Benjamin_Boenisch/breakpilot-lehrer
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(ocr-cells): align cell bboxes exactly to column/row coordinates
Some checks failed
CI / go-lint (push) Has been skipped
Details
CI / python-lint (push) Has been skipped
Details
CI / nodejs-lint (push) Has been skipped
Details
CI / test-go-school (push) Successful in 25s
Details
CI / test-go-edu-search (push) Successful in 26s
Details
CI / test-python-klausur (push) Failing after 1m49s
Details
CI / test-python-agent-core (push) Successful in 17s
Details
CI / test-nodejs-website (push) Successful in 17s
Details

Browse Source 
Decouple display bbox from OCR crop region. Display bbox now uses exact
col.x/row.y/col.width/row.height (no padding), so adjacent cells touch
without gaps. OCR crop keeps 4px internal padding for edge character
detection.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Benjamin Admin
2026-03-03 09:21:56 +01:00
parent 4d428980c1
commit 75bca1f02d
1 changed files with 16 additions and 15 deletions
Download Patch File Download Diff File Expand all files Collapse all files

31
klausur-service/backend/cv_vocab_pipeline.py
View File

@@ -3807,19 +3807,20 @@ def _ocr_single_cell(
preassigned_words: Optional[List[Dict]] = None,
) -> Dict[str, Any]:
"""Populate a single cell (column x row intersection) via word lookup."""
pad = 8 # pixels
# Display bbox: exact column × row intersection (no padding)
disp_x = col.x
disp_y = row.y
disp_w = col.width
disp_h = row.height
# OCR crop: slightly wider to catch edge characters (internal only)
pad = 4
cell_x = max(0, col.x - pad)
cell_y = max(0, row.y - pad)
cell_w = col.width + 2 * pad
cell_h = row.height + 2 * pad
cell_w = min(col.width + 2 * pad, img_w - cell_x)
cell_h = min(row.height + 2 * pad, img_h - cell_y)
# Clamp to image bounds
if cell_x + cell_w > img_w:
cell_w = img_w - cell_x
if cell_y + cell_h > img_h:
cell_h = img_h - cell_y
if cell_w <= 0 or cell_h <= 0:
if disp_w <= 0 or disp_h <= 0:
return {
'cell_id': f"R{row_idx:02d}_C{col_idx}",
'row_index': row_idx,
@@ -3929,12 +3930,12 @@ def _ocr_single_cell(
'col_type': col.type,
'text': text,
'confidence': avg_conf,
'bbox_px': {'x': cell_x, 'y': cell_y, 'w': cell_w, 'h': cell_h},
'bbox_px': {'x': disp_x, 'y': disp_y, 'w': disp_w, 'h': disp_h},
'bbox_pct': {
'x': round(cell_x / img_w * 100, 2),
'y': round(cell_y / img_h * 100, 2),
'w': round(cell_w / img_w * 100, 2),
'h': round(cell_h / img_h * 100, 2),
'x': round(disp_x / img_w * 100, 2),
'y': round(disp_y / img_h * 100, 2),
'w': round(disp_w / img_w * 100, 2),
'h': round(disp_h / img_h * 100, 2),
},
'ocr_engine': used_engine,
}