From 18ea0fb7f9a69a3b682f0111fa01a66970ebb1fd Mon Sep 17 00:00:00 2001 From: Dexterity <173429049+Dexterity104@users.noreply.github.com> Date: Wed, 22 Jul 2026 02:28:44 -0400 Subject: [PATCH] fix(deepdoc): prevent figure and equation crops from merging on scientific PDFs (#15873) ### What problem does this PR solve? Closes #15872 On pages that contain both a textless equation and a figure, the layout recognizer could merge the two unrelated regions into a single cropped image with concatenated captions. This shows up most often on scientific and technical PDFs. The cause is a `layoutno` namespace collision in `deepdoc/vision/layout_recognizer.py`. Text-overlapping boxes are tagged per type by `findLayout`: figures become `figure-{ii}` using the figure-only index, and equations become `equation-{ii}` using the equation-only index. The fallback loop that handles textless regions, however, indexes the combined figure plus equation list and always uses a `figure` prefix. Because the two paths use different index spaces and prefixes, a page laid out as `[textless equation, figure with text]` produces two boxes tagged `figure-0`. `_extract_table_figure` in `deepdoc/parser/pdf_parser.py` buckets boxes by `f"{page}-{layoutno}"`, so both fall into the same `page-figure-0` bucket, and `cropout` stitches the disjoint regions into one image. **Fix** The fallback loop now iterates per type, indexes within each type's own list, and reuses the type as the `layoutno` prefix (`figure-{i}` or `equation-{i}`). This matches the namespace that `findLayout` already assigns to text-overlapping boxes. Since the `visited` flag is shared by reference, each layout is tagged by exactly one path, so per-type indices stay collision free. Textless equations now land under `equation-N`, consistent with text-overlapping equations, instead of the old `figure-N`. The same fix is applied to `AscendLayoutRecognizer`, which shared the identical defect. Downstream consumers of `layoutno` were checked and all treat it as an opaque equality or bucketing key, so no other code paths are affected. When a page has no equations, the combined index equals the figure-only index and the output is unchanged. ### Type of change - [x] Bug Fix (non-breaking change which fixes an issue) --- deepdoc/vision/layout_recognizer.py | 53 ++++++++++++++++++----------- 1 file changed, 33 insertions(+), 20 deletions(-) diff --git a/deepdoc/vision/layout_recognizer.py b/deepdoc/vision/layout_recognizer.py index 21bc1f4bad..e71b6a654c 100644 --- a/deepdoc/vision/layout_recognizer.py +++ b/deepdoc/vision/layout_recognizer.py @@ -136,16 +136,23 @@ class LayoutRecognizer(Recognizer): for lt in ["footer", "header", "reference", "figure caption", "table caption", "title", "table", "text", "figure", "equation"]: findLayout(lt) - # add box to figure layouts which has not text box - for i, lt in enumerate([lt for lt in lts if lt["type"] in ["figure", "equation"]]): - if lt.get("visited"): - continue - lt = deepcopy(lt) - del lt["type"] - lt["text"] = "" - lt["layout_type"] = "figure" - lt["layoutno"] = f"figure-{i}" - bxs.append(lt) + # add box to figure/equation layouts which have no text box. + # Index within each type's own list and keep the type as the layoutno + # prefix so these match the namespace findLayout() assigns to + # text-overlapping boxes (figure-N vs equation-N). Using a combined + # figure+equation index with a fixed "figure" prefix would collide + # with figure-N tags from findLayout and merge unrelated regions. + for ty in ["figure", "equation"]: + for i, lt in enumerate([lt for lt in lts if lt["type"] == ty]): + if lt.get("visited"): + continue + lt = deepcopy(lt) + lt.pop("type", None) + lt["text"] = "" + lt["layout_type"] = "figure" + lt["layoutno"] = f"{ty}-{i}" + logging.debug(f"Created placeholder box {lt['layoutno']} for textless {ty} region") + bxs.append(lt) boxes.extend(bxs) @@ -437,16 +444,22 @@ class AscendLayoutRecognizer(Recognizer): for ty in ["footer", "header", "reference", "figure caption", "table caption", "title", "table", "text", "figure", "equation"]: _tag_layout(ty) - figs = [lt for lt in lts if lt["type"] in ["figure", "equation"]] - for i, lt in enumerate(figs): - if lt.get("visited"): - continue - lt = deepcopy(lt) - lt.pop("type", None) - lt["text"] = "" - lt["layout_type"] = "figure" - lt["layoutno"] = f"figure-{i}" - bxs.append(lt) + # Index within each type's own list and keep the type as the layoutno + # prefix so these match the namespace _tag_layout() assigns to + # text-overlapping boxes (figure-N vs equation-N). Using a combined + # figure+equation index with a fixed "figure" prefix would collide + # with figure-N tags from _tag_layout and merge unrelated regions. + for ty in ["figure", "equation"]: + for i, lt in enumerate([lt for lt in lts if lt["type"] == ty]): + if lt.get("visited"): + continue + lt = deepcopy(lt) + lt.pop("type", None) + lt["text"] = "" + lt["layout_type"] = "figure" + lt["layoutno"] = f"{ty}-{i}" + logging.debug(f"Created placeholder box {lt['layoutno']} for textless {ty} region") + bxs.append(lt) boxes_out.extend(bxs)