• # Gros titre à ...

    Posté par . En réponse à la dépêche Tesseract-OCR. Évalué à 8.

    Je préfère l'information originale, cf. README:

    History:
    The engine was developed at Hewlett Packard Laboratories Bristol and at Hewlett Packard Co, Greeley Colorado between 1985 and 1994, with some more changes made in 1996 to port to Windows, and some C++izing in 1998. A lot of the code was written in C, and then some more was written in C++. Since then all the code has been converted to at least compile with a C++ compiler. Currently it builds under Linux with gcc2.95 and under Windows with VC++6. The C++ code makes heavy use of a list system using macros. This predates stl, was portable before stl, and is more efficent than stl lists, but has the big negative that if you do get a segmentation violation, it is hard to debug. Another "feature" of the C/C++ split is that the C++ data structures get converted to C data structures to call the low-level C code. This is ugly, and the C++izing of the C code is a step towards eliminating the conversion, but it has not happened yet.

    About the Engine:
    This code is a raw OCR engine. It has NO PAGE LAYOUT ANALYSIS, NO OUTPUT FORMATTING, and NO UI. It can only process an image of a single column and create text from it. It can detect fixed pitch vs proportional text. Having said that, in 1995, this engine was in the top 3 in terms of character accuracy, and it compiles and runs on both Linux and Windows. Another current limitation is that it only recognizes English and its character set is only US-ASCII. Training code IS included in the open source release however, and will be included in a future release.


    Mon expérience sur les programmes datant de 20 ans: l'expertise sur le sujet (de l'expérience et de la littérature) est plus importante que le code source. Un logiciel a un cycle de vie comme les humains. Passé un certain temps il faut repartir sur des bases nouvelles. Cela dit disposer du code source c'est mieux que rien.

    Gocr donne aussi de bons résultats. Son gros défaut provient du fait qu'il n'est pas capable d'apprendre une police, par exemple les ligatures des CM de TeX. D'après le README de Tesseract-OCR il y aurait un apprentissage.

    La route est longue avant d'avoir un moteur ocr complet. Mais avec les millions de Google on peut rêver ...