degradation of chemical structures identification in pdfs
Brought to you by:
igor_filippov
Performance of chemical structures identification in PDFs seems to be a bit degraded with moving 2.0.1 -> 2.1.0.
osra -c -e -p -g -f smi -o test -w test.csv jbc_2013_baba.pdf
2.0.1: finds~30 structures with 2-3 false postitives (exact numbers depends on ocrad version with which it's linked).
2.1.0: finds only 2 structures, no false positives though.
Indeed. The structure images are a bit on the small size in the attached pdf but I will add them to the regression test set. Thank you.