Tom, Is it not possible if data stored in RAM exceeds 1GB moved or transfer to Swap pragmatically OR data stored in RAM moved *every seconds* to the Swap and cleared up RAM for next processing the data. - I presumed that this can be done by suitable coding in the relevant source code.in python. I am of view majority of users may not have higher end of computer having more than 4GB - resulting discourage them for usage of ocropus project. i searched for such type of computer but not available - all depends on motherboards..
If you feel ocropus project is meant only the use of *large scale machine learning* for addressing problems in document analysis. and NOT for any lang OCR purpose( i.e. small scale) by users. Kindly confirm. With warmest regards, -sriranga(79yrs) On Fri, Oct 5, 2012 at 6:42 PM, Sriranga(78yrsold) <[email protected]>wrote: > Tom, > As suggested by you that training requires a lot of memory, I am thinking > whether it is feasible to create swap area to 20GB in addition to existing > 4GB RAM - which is maximum for my Dell machine. However I find program does > not utilize the swap area effectively - for which suitable coding is > required I guess - ultimately solve the problem of memory error. At present > swap area is 16GB. > Alternatively whether external HDD or internal one HDD can be utilized as > combined RAM+swap? > With Warmest Regards, > -sriranga(79yrs) > > > On Fri, Oct 5, 2012 at 2:48 PM, Tom <[email protected]> wrote: > >> wed, I'm sorry I couldn't help you more. , am not aware Of any >> memoryleaks, but as I indicated, training requires a lot of memory, We >> Usually use 8-16GB of memory for training. >> >> Tom >> On Oct 4, 2012 9:25 AM, "Sriranga(78yrsold)" <[email protected]> >> wrote: >> >>> >>> My computer Dell optiplex 330(System Type X86-based PC) >>> which has Processor Intel(R)core(TM)DuoCPU [email protected] 2.53GHz OS >>> win XPwith Sp3. RAM 4GB(recently added more ram) Ubuntu 12.04(32bit) and >>> also 12.10.(both has swap of 16GB) It appears that my machine does not >>> support Ocropus project which requires higher capacity computer.. >>> >>> After several experiments with python 0.6 it is observed very often >>> memory error will displayed. I feel there is leakage of memory in relevant >>> python source codes of 0.6 more over some of py programs does not support >>> kannada script except during run-test and run-box-training. Since i am not >>> python programmer nor developer. i find difficult to pursue the kannada OCR >>> project under Ocropus Project due to support not received from any one >>> except from you but you are too busy in release of next stage of version. >>> As such I am frustrated in kannada OCR project of Ocropus - decided to >>> discontinue. >>> I am always ready to furnish Kannada text, tif or png with its box file >>> generated in tesseract-ocr. to any python programmer or developer for >>> research purpose at any time.and also undertake beta-testing/feedback of >>> py programs for kannada at any stage, if any, received from you. >>> >>> Thanks for the help rendered to me from time to time. >>> With warmest Regards, >>> -sriranga(79yrs) >>> >>> >>> >>> On Thu, Oct 4, 2012 at 12:23 PM, Tom <[email protected]> wrote: >>> >>>> Yes basically. The "scale" is the estimated size of the character >>>> x-height in pixels, and your is too small. Your characters are set in a >>>> 3pt font on a 300dpi page. >>>> >>>> Tom >>>> >>>> On Wednesday, October 3, 2012 8:23:18 AM UTC-7, JR wrote: >>>>> >>>>> What does "scale (6.9282) less than --minscale; skipping" really >>>>> mean? is it the resolution of the image? DPI? >>>>> >>>>> I too have this issue. Curious to know what the issue is? >>>>> >>>>> >>>>> On Friday, August 24, 2012 6:12:36 AM UTC-4, carolin wrote: >>>>>> >>>>>> Hi, >>>>>> >>>>>> we installed ocropus 0.6 on an virtual Ubuntu 12.04 maschine >>>>>> according to the instructrions on your web page and got the following >>>>>> assertion on one of our files when using the ocropus-recognize-book >>>>>> script: >>>>>> >>>>>> carolin@carolin-VirtualBox:~/**ocrinput$ ocropus-recognize-book >>>>>> testbild2.png -o output.html >>>>>> book directory ./_book-003266 >>>>>> testbild2.png -> ./_book-003266/0001.png >>>>>> >>>>>> === preprocess >>>>>> >>>>>> # ocropus-nlbin ./_book-003266/????.png >>>>>> === ./_book-003266/0001.png 1 >>>>>> flattening >>>>>> estimating skew angle >>>>>> estimating thresholds >>>>>> rescaling >>>>>> ./_book-003266/0001.png lo-hi (0.05 1.00) angle 0.0 >>>>>> writing >>>>>> >>>>>> === page segmentation >>>>>> >>>>>> # ocropus-gpageseg ./_book-003266/????.bin.png >>>>>> ./_book-003266/0001.bin.png >>>>>> ./_book-003266/0001.bin.png: scale (6.9282) less than --minscale; >>>>>> skipping >>>>>> >>>>>> === line recognition >>>>>> >>>>>> # ocropus-lattices ./_book-003266/????/??????.**bin.png >>>>>> Traceback (most recent call last): >>>>>> File "/usr/local/bin/ocropus-**lattices", line 56, in <module> >>>>>> args.files = ocrolib.glob_all(args.files) >>>>>> File "/usr/local/lib/python2.7/**dist-packages/ocrolib/**toplevel.py", >>>>>> line 204, in argument_checks >>>>>> result = f(*args,**kw) >>>>>> File "/usr/local/lib/python2.7/**dist-packages/ocrolib/common.**py", >>>>>> line 509, in glob_all >>>>>> raise Exception("%s: expansion did not yield any files"%arg) >>>>>> Exception: ./_book-003266/????/??????.**bin.png: expansion did not >>>>>> yield any files >>>>>> Traceback (most recent call last): >>>>>> File "/usr/local/bin/ocropus-**recognize-book", line 79, in >>>>>> <module> >>>>>> run(args.linerec,book+"/????/?**?????.bin.png",m=args.model) >>>>>> File "/usr/local/bin/ocropus-**recognize-book", line 57, in run >>>>>> assert subprocess.call(args)==0 >>>>>> AssertionError >>>>>> >>>>>> We attached the sample file which we used. Is there any obvious thing >>>>>> that we are doing wrong? >>>>>> >>>>>> Best, >>>>>> >>>>>> Carolin >>>>>> >>>>> -- >>>> You received this message because you are subscribed to the Google >>>> Groups "ocropus" group. >>>> To post to this group, send email to [email protected]. >>>> To unsubscribe from this group, send email to >>>> [email protected]. >>>> To view this discussion on the web visit >>>> https://groups.google.com/d/msg/ocropus/-/c_q4jBZOA8UJ. >>>> >>>> For more options, visit https://groups.google.com/groups/opt_out. >>>> >>>> >>>> >>> >>> > -- You received this message because you are subscribed to the Google Groups "ocropus" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit https://groups.google.com/groups/opt_out.
