CLiDE: Chemical Literature Data Extraction

Validation

Reference:

1. Aniko T. Valko and A. Peter Johnson: "CLiDE Pro: The Latest Generation of CLiDE, a Tool for Optical Chemical Structure Recognition"
   J. Chem. Inf. Model., 2009, 49 (4), pp 780-787
   DOI: 10.1021/ci800449t

The following test set (as reported in Ref. 1) was processed with CLiDE Pro in "batch mode". The images were processed and saved automatically. No manual correction of the results was applied (although CLiDE gives opportunity for this with various editors). The input and output files along with the corresponding processing times and success rate for the test set  are provided below:

Below you'll find OLD¹ data & results with CLiDE v2.1 - from the 20th century :-)  
(CLiDE Lite and Full v 2.1 were developed from 1991-1999)

Note-1:  These older datasets and results are kept only becauase the scientific community used these datasets as STANDARD / reference validation sets - i.e. the CLiDE development team under the guidance of Prof. A. Peter Johnson showed leadership in validating Chemical OCR methods from the early days.

HW	# images	Input files	Output	Proc. Time
366 MHz PII lap-top with 160 MB memory	14	Input: 14 images in Windows bitmap file format were processed	Output 1: saved in MOL file format Output 2: saved in ChemDraw file format Output 3 : saved in CLiDE file format	30 sec
900 MHz ATHLON PC with 768 MB memory				8 sec
366 MHz PII lap-top with 160 MB memory	96	Input: 96 images in Windows bitmap file format were processed	Output 1: saved in MOL file format Output 2: saved in ChemDraw file format Output 3: saved in CLiDE file format	650 sec (ca. 11 minutes)
900 MHz ATHLON PC with 768 MB memory				230 sec (ca. 4 minutes)

Some of the images in the validation set can be viewed below (click on the image to see its processing details):