{"id":105302,"date":"2022-10-13T15:36:48","date_gmt":"2022-10-13T13:36:48","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/otkryt-kod-sistemy-mashinnogo-obucheniya-dlya-generaczii-realistichnyh-dvizhenij-cheloveka"},"modified":"2022-10-13T15:36:48","modified_gmt":"2022-10-13T13:36:48","slug":"otkryt-kod-sistemy-mashinnogo-obucheniya-dlya-generaczii-realistichnyh-dvizhenij-cheloveka","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sq\/blog\/novosti-interneta\/otkryt-kod-sistemy-mashinnogo-obucheniya-dlya-generaczii-realistichnyh-dvizhenij-cheloveka","title":{"rendered":"I \u00ebsht\u00eb hapur kodi i sistemit t\u00eb m\u00ebsimit t\u00eb makineris\u00eb p\u00ebr gjenerimin e l\u00ebvizjeve realiste t\u00eb njeriut.","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
Nj\u00eb grup hulumtuesish nga Universiteti i Tel Avivit ka publikuar kodin burimor t\u00eb lidhur me sistemin e m\u00ebsimit t\u00eb makinerive MDM (Motion Diffusion Model), i cili lejon gjenerimin e l\u00ebvizjeve realiste t\u00eb njer\u00ebzve. Kodi \u00ebsht\u00eb shkruar n\u00eb gjuh\u00ebn Python duke p\u00ebrdorur korniz\u00ebn PyTorch dhe shp\u00ebrndahet n\u00ebn licenc\u00ebn MIT. P\u00ebr t\u00eb kryer eksperimente, mund t\u00eb p\u00ebrdoren modele t\u00eb gatshme, si dhe t\u00eb kryhet st\u00ebrvitja e modeleve vet\u00eb me ndihm\u00ebn e skripteve t\u00eb ofruara, p\u00ebr shembull, duke p\u00ebrdorur koleksionin e imazheve tre-dimensionale t\u00eb njeriut HumanML3D. P\u00ebr t\u00eb trajnuar sistemin, nevojitet nj\u00eb GPU me mb\u00ebshtetje CUDA. <\/p>\n
Aplikimi i mund\u00ebsive tradicionale p\u00ebr animimin e l\u00ebvizjeve t\u00eb njer\u00ebzve \u00ebsht\u00eb i v\u00ebshtir\u00eb p\u00ebr shkak t\u00eb komplikimeve q\u00eb lidhen me ndryshueshm\u00ebrin\u00eb e madhe t\u00eb l\u00ebvizjeve t\u00eb mundshme dhe v\u00ebshtir\u00ebsin\u00eb e p\u00ebrshkrimit t\u00eb tyre formal, si dhe p\u00ebr shkak t\u00eb ndjeshm\u00ebris\u00eb s\u00eb madhe t\u00eb perceptimit njer\u00ebzor ndaj l\u00ebvizjeve t\u00eb paq\u00ebndrueshme. P\u00ebrpjekjet e m\u00ebparshme p\u00ebr t\u00eb p\u00ebrdorur modelet gjenerative t\u00eb m\u00ebsimit t\u00eb makinerive pat\u00ebn probleme me cil\u00ebsin\u00eb dhe shprehshm\u00ebrin\u00eb e kufizuar. <\/p>\n
N\u00eb sistemin e propozuar \u00ebsht\u00eb b\u00ebr\u00eb nj\u00eb p\u00ebrpjekje p\u00ebr t\u00eb p\u00ebrdorur modelet difuzive p\u00ebr gjenerimin e l\u00ebvizjeve, t\u00eb cilat n\u00eb thelb jan\u00eb m\u00eb t\u00eb p\u00ebrshtatshme p\u00ebr simuluar l\u00ebvizjet njer\u00ebzore, por nuk jan\u00eb pa disavantazhe, si\u00e7 jan\u00eb k\u00ebrkesat e larta p\u00ebr burime kompjuterike dhe kompleksiteti n\u00eb menaxhimin e tyre. P\u00ebr t\u00eb minimizuar disavantazhet e modeleve difuzive, n\u00eb MDM \u00ebsht\u00eb angazhuar nj\u00eb rrjet nervor me arkitektur\u00ebn \"transformer\" dhe parashikimi i mostrat (sample) n\u00eb vend t\u00eb parashikimit t\u00eb zhurm\u00ebs n\u00eb \u00e7do hap, \u00e7ka e thjeshton parandalimin e anomali-ve, si humbja e kontaktit midis sip\u00ebrfaqes dhe k\u00ebmb\u00ebs. <\/p>\n
P\u00ebr menaxhimin e gjenerimit \u00ebsht\u00eb parashikuar mund\u00ebsia e aplikimit t\u00eb p\u00ebrshkrimit tekstual t\u00eb veprimit n\u00eb gjuh\u00ebn natyrore (p.sh. \"nj\u00eb njeri po ec\u00ebn p\u00ebrpara dhe po p\u00ebrkulet p\u00ebr t\u00eb ngritur di\u00e7ka nga toka\") ose p\u00ebrdorimi i veprimeve standarde, si \"vrapim\" dhe \"k\u00ebrcime\". Sistemi gjithashtu mund t\u00eb p\u00ebrdoret p\u00ebr t\u00eb redaktuar l\u00ebvizjet dhe p\u00ebr t\u00eb plot\u00ebsuar detaje t\u00eb humbura. K\u00ebrkuesit kan\u00eb kryer teste, ku pjes\u00ebmarr\u00ebsve ju ofruan t\u00eb zgjidhnin nga disa opsione rezultatin m\u00eb t\u00eb mir\u00eb - n\u00eb 42% t\u00eb rasteve, njer\u00ebzit preferuan l\u00ebvizjet e sintetizuara n\u00eb vend t\u00eb atyre reale.
![\"Luaj](\"https:\/\/img.youtube.com\/vi\/rVkIDj5wgjs\/hqdefault.jpg\")
\r\n