%0 Journal Article
%9 ACL : Articles dans des revues avec comité de lecture non répertoriées par l'AERES
%A Le Bourgeois, T.
%A Bonnet, P.
%A Edelin, C.
%A Grard, P.
%A Prosperi, J.
%A Théveny, F.
%A Barthélémy, Daniel
%T L'identification des adventices assistée par ordinateur avec le système IDAO
%D 2008
%L PAR00007455
%G FRE
%J Innovations Agronomiques
%K Identification assisted by computer ; Identikit ; IDAO ; Weed
%M HAL:halsde-00344614
%P 167-175
%U https://www.documentation.ird.fr/hor/PAR00007455
%V 3
%W Horizon (IRD)
%X Identification of crop weeds is essential to get the information needed for elaborating efficient control methods. Non specialised people had difficulties to do this identification with classical tools, such as floras or field guides (too technical, unsuitable for seedlings or partial samples, process difficult to follow...). That brought us to develop a new system for plant recognition assisted by computer that was called IDAO (IDentification Assistée par Ordinateur). This software has the distinctive feature to use a graphical identification system by identikit. This identikit allows the user to build the image of the plant from traits freely chosen according to the specimen or to the user. It tolerates observation errors or polymorphism. Species are listed by their probability of similarity with the identikit. Descriptions, illustrations and information (biology, ecology, control...) are available at any time in local or online Html pages. These descriptive files can be regularly updated on the Web site. IDAO is a multilingual and multiplatform system. It can be used on PC (from cdrom or downloaded) or directly in the field on ultra mobile computer. Several applications have been published on weed floras of different cropping systems (rice, cotton, food crops, sugarcane...) and for different world areas (Africa, Asia, India, Indian Ocean), and also for other kinds of plants (trees, orchids...). The IDAO system will evolve during the Pl@ntnet project that will start in early 2009. IDAO will be available as free software on an Internet platform, for every body can develop by himself or under collaboration new applications available for all the user community. This identification system will be linked to an automatic recognition tool, using image analysis.
%$ 076 ; 122