it's 3.31.3 release day, upload a tarball and show that our software is more stable than this table 彡┻━┻

Commit d1da8cd2 authored by Jonas Wloka's avatar Jonas Wloka Committed by Jerome Flesch

Filter out boxes that start at (0, 0)

Tesseract returns way too large boxes that cover the whole page, mostly
containing only a single special character. All of these boxes (in my
tests) have coordinate 0 0.

This filters out all of these boxes.
parent b65b13f7
......@@ -251,10 +251,14 @@ class JobPageBoxesLoader(Job):
boxes = set()
for line in line_boxes:
for word in line.word_boxes:
if word.content.strip() == "":
if word.content.strip() == "" or (word.position[0][0] == 0
and word.position[0][1] == 0):
# XXX(Jflesch): Tesseract 3.03 (hOCR) returns big and
# empty word boxes sometimes (just a single space
# inside). They often match images, but not always.
# XXX(jowlo): Tesseract returns large boxes containing
# single letters (mostly special chars) that cover the
# whole page. All of these start at (0 0)
continue
boxes.add(word)
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment