Skip to content
Snippets Groups Projects

Repository graph

You can move around the graph by using the arrow keys.
Select Git revision
  • master default protected
1 result
Created with Raphaël 2.2.01Feb7Nov26Oct231110629Aug1812Jul527Apr329Mar2813Oct1Mar10Jan27May252418Feb171632122Jan212010Dec820Aug17Jul1625May15Mar17Feb111Jan98543222Dec528Nov2726330Oct271413Sep31Aug30201716124Jul20Jun18151413einheitliche APImastermasterupdated modelbetter formattingDocstring Documentationrefactoringget Namelist from clientbetter preprocessor for training dataable to get all student infosUmlaute beachtenadded lexicon search for Name recognitionupdated for real exam exampleget top n beamssend all Matr. numberssplit corpus into Vor und Nachnameadd webserveradded model and corpusadded link to demoMerge pull request #157 from alexeevrevan/masterFixed Bugs and Added Train Loss Dump Featureadded link to repo that uses this modelignore empty lines and comment lines in IAM dataset filein validation mode use charset from trained model and not from dataset, see issue #127png images: white instead of transparent bg to handle dark modereadme, docstrings, and other small changesready for first line-level training runsuse correct batch sizereworked preprocessorstarted multi word trainingallow dynamic width of imagechanged code to python coding stylerm old faq sectionword beam search: switched from TF custom op to Python package, so it can also be used in Windows nowcleanup: remove analyze code, and only keep code really relevant for HTRsingle option for decoderslog both word acc and cer for valsetchanged data augdata aug: dont move img around if it is already larger than target imgall model files are now in the uploaded zipdata aug: random background, write summary of CERs while trainingtest: data aug=true