{"id":33918,"date":"2024-04-24T13:35:47","date_gmt":"2024-04-24T11:35:47","guid":{"rendered":"https:\/\/www.westcam-group.com\/services\/vyvoj-strojoveho-uceni-umele-inteligence\/"},"modified":"2025-07-16T09:43:11","modified_gmt":"2025-07-16T07:43:11","slug":"vyvoj-strojoveho-uceni-umele-inteligence","status":"publish","type":"service","link":"https:\/\/www.westcam-group.com\/cs\/sluzby\/reseni-pro-zobrazovani\/strojove-uceni\/vyvoj-strojoveho-uceni-umele-inteligence\/","title":{"rendered":"V\u00fdvoj strojov\u00e9ho u\u010den\u00ed \/ um\u011bl\u00e9 inteligence"},"content":{"rendered":"\n
\n
\n
\n
M\u00edsto klasick\u00e9ho zpracov\u00e1n\u00ed obrazu a „inteligentn\u00edch“ senzor\u016f plug & play <\/div>\n\n\n\n

Kontrola kvality pomoc\u00ed um\u011bl\u00e9 inteligence (AI)<\/h1>\n\n\n\n

Intenzivn\u011b se zab\u00fdv\u00e1me strojov\u00fdm u\u010den\u00edm – podoblast\u00ed um\u011bl\u00e9 inteligence.<\/strong><\/p>\n\n\n\n

Na\u0161e strojov\u00e9 u\u010den\u00ed<\/strong> vych\u00e1z\u00ed z oblasti klasick\u00e9ho zpracov\u00e1n\u00ed obrazu a tak\u00e9<\/strong> se to\u010d\u00ed kolem zpracov\u00e1n\u00ed obrazov\u00fdch dat. <\/p>\n\n\n\n

M\u016f\u017eeme tak\u00e9 realizovat slo\u017eit\u011bj\u0161\u00ed \u00falohy, kter\u00e9 vy\u017eaduj\u00ed kombinaci klasick\u00fdch algoritm\u016f a strojov\u00e9ho u\u010den\u00ed nebo kombinaci r\u016fzn\u00fdch zdroj\u016f dat (anal\u00fdza n\u011bkolika sn\u00edmk\u016f objektu, integrace dal\u0161\u00edch dat ze senzor\u016f nebo vstup\u016f od u\u017eivatele).<\/p>\n\n\n\n

\n
Poptejte nyn\u00ed<\/a><\/div>\n<\/div>\n<\/div>\n\n\n\n
\n
\"Kontrola<\/figure>\n\n\n\n
\n
Nejd\u016fle\u017eit\u011bj\u0161\u00ed informace<\/div>\n\n\n\n
    \n
  • Klasifikace<\/li>\n\n\n\n
  • Multimod\u00e1ln\u00ed zpracov\u00e1n\u00ed dat<\/li>\n\n\n\n
  • Detekce a lokalizace objekt\u016f<\/li>\n\n\n\n
  • Detekce anom\u00e1li\u00ed<\/li>\n<\/ul>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n\n\n\n
    \n
    \"\"<\/figure>
    \n

    Klasifikace<\/h3>\n\n\n\n
      \n
    • Diferenciace do libovoln\u00fdch t\u0159\u00edd (nap\u0159. [dobr\u00fd , \u0161patn\u00fd] nebo [dobr\u00fd , typ chyby 1 , typ chyby 2]).<\/li>\n\n\n\n
    • Tr\u00e9nov\u00e1n\u00ed na z\u00e1klad\u011b velk\u00e9ho mno\u017estv\u00ed tr\u00e9novac\u00edch dat pro v\u0161echny t\u0159\u00eddy.<\/span><\/li>\n\n\n\n
    • Mo\u017enost detekce velmi jemn\u00fdch rozd\u00edl\u016f.<\/li>\n<\/ul>\n<\/div><\/div>\n\n\n\n
      \n

      Multimod\u00e1ln\u00ed zpracov\u00e1n\u00ed dat („roz\u0161\u00ed\u0159en\u00e1 klasifikace“)<\/h3>\n\n\n\n
        \n
      • Diferenciace do libovoln\u00fdch t\u0159\u00edd (nap\u0159. [optim\u00e1ln\u00ed kvalita , pr\u016fm\u011brn\u00e1 kvalita , zmetky]).<\/li>\n\n\n\n
      • Tr\u00e9nov\u00e1n\u00ed na z\u00e1klad\u011b velk\u00e9ho mno\u017estv\u00ed tr\u00e9novac\u00edch dat pro v\u0161echny t\u0159\u00eddy.<\/span><\/li>\n\n\n\n
      • Kombinuj\u00ed se vstupn\u00ed data z r\u016fzn\u00fdch zdroj\u016f, nap\u0159. n\u011bkolik sn\u00edmk\u016f z kamery, dal\u0161\u00ed informace o v\u00fdrobku, jako je barva, tvar nebo hmotnost, nebo \u00fadaje zadan\u00e9 u\u017eivatelem.<\/span><\/li>\n\n\n\n
      • Mo\u017enost kombinace r\u016fzn\u00fdch algoritm\u016f.<\/span><\/li>\n\n\n\n
      • Rozpozn\u00e1n\u00ed rozd\u00edl\u016f, kter\u00e9 je mo\u017en\u00e9 pouze na z\u00e1klad\u011b kombinace r\u016fzn\u00fdch \u00fadaj\u016f. Je mo\u017en\u00e1 kontrola cel\u00e9ho v\u00fdrobku (nap\u0159. cel\u00e9ho povrchu prost\u0159ednictv\u00edm n\u011bkolika sn\u00edmk\u016f, kter\u00e9 se vyhodnocuj\u00ed spole\u010dn\u011b). <\/li>\n<\/ul>\n<\/div>
        \"\"<\/figure><\/div>\n\n\n\n
        \"\"<\/figure>
        \n

        Rozpozn\u00e1v\u00e1n\u00ed a detekce<\/h3>\n\n\n\n

        Detekce a lokalizace objekt\u016f <\/span><\/strong><\/p>\n\n\n\n

          \n
        • Rozpozn\u00e1v\u00e1n\u00ed p\u0159edm\u011bt\u016f v\u010detn\u011b ur\u010dov\u00e1n\u00ed jejich polohy.<\/span><\/li>\n\n\n\n
        • Tr\u00e9nov\u00e1n\u00ed na z\u00e1klad\u011b tr\u00e9ninkov\u00fdch dat v\u0161ech objekt\u016f.<\/span><\/li>\n\n\n\n
        • To umo\u017e\u0148uje mimo jin\u00e9 kontrolu p\u0159\u00edtomnosti nebo po\u010d\u00edt\u00e1n\u00ed objekt\u016f.<\/li>\n<\/ul>\n\n\n\n

          Detekce anom\u00e1li\u00ed<\/strong><\/p>\n\n\n\n

            \n
          • Rozli\u0161en\u00ed mezi „norm\u00e1ln\u00edmi“ a „anom\u00e1ln\u00edmi“ daty.<\/li>\n\n\n\n
          • Tr\u00e9nink se obvykle prov\u00e1d\u00ed s mal\u00fdm mno\u017estv\u00edm norm\u00e1ln\u00edch a abnorm\u00e1ln\u00edch dat. Pokud to nen\u00ed mo\u017en\u00e9 jinak, je mo\u017en\u00e9 tr\u00e9novat na z\u00e1klad\u011b 100 % norm\u00e1ln\u00edch p\u0159\u00edklad\u016f. <\/span><\/li>\n\n\n\n
          • V z\u00e1vislosti na stupni anom\u00e1li\u00ed je mo\u017en\u00e1 st\u0159edn\u00ed a\u017e dobr\u00e1 p\u0159esnost detekce.<\/li>\n<\/ul>\n<\/div><\/div>\n<\/div>\n\n\n\n
            \n

            \u010casto kladen\u00e9 ot\u00e1zky (FAQ)<\/h3>\n\n\n\n