Uhlolojikelele
Amanethiwekhi Esalela (ResNets) amanethiwekhi ajulile e-neural angeza 'ukweqa ukuxhumana' okuvumela izendlalelo zifunde ukulungiswa okuncane esikhundleni sokuguqulwa okugcwele. Leli qhinga elilula lenze kwaba nokwenzeka ukuqeqesha amanethiwekhi amakhulu ezendlalelo ezijulile, okuvusa ukunemba kokubonwa kwesithombe.
Amanethiwekhi Esalela awokugeleza kokusebenza kombono wekhompuyutha okuhumusha noma okukhiqiza imidiya ebonakalayo ukuze ihlaziywe, isebenze, futhi isungule.
I-Deep Dive
Ngaphambi kwe-ResNets, ukupakisha izendlalelo eziningi ngokuxakayo kwenza amanethiwekhi enze kabi kakhulu, ngisho nakudatha yokuqeqeshwa, inkinga ebizwa ngokuthi ukuwohloka. Ngo-2015, Microsoft abacwaningi u-Kaiming He kanye nozakwabo bethula ibhulokhi eyinsalela: esikhundleni sokucela inqwaba yezendlalelo ukuthi ikhiqize okukhiphayo okungu-H(x) ngokuqondile, bayivumela ifunde okusele F(x) = H(x) - x, base bengeza okokufaka koqobo x emuva ngesinqamuleli. Uma ungqimba lungadingeki, lungavele lufunde ukungenzi lutho (F(x) = 0). I-ResNet-152 iwine umncintiswano we-ImageNet wango-2015 ngephutha eliphezulu-5 lamaphesenti angaba ngu-3.6, yehlula izilinganiso zezinga labantu, futhi ukwakheka kwayo kwaba umgogodla oyisisekelo wokutholwa, ukuhlukaniswa, nokuthatha izithombe zezokwelapha.
I-Technical Insight
Uxhumano lwe-skip lushintsha umsebenzi webhulokhi ngalinye ube ngu-y = F(x) + x. Ngesikhathi sokusakazwa ngemuva, i-gradient igeleza kusinqamuleli sobunikazi singashintshiwe, ngakho-ke asikwazi ukunyamalala siye eduze kweziro ngisho nakumakhulu ezendlalelo. Lokhu kugcina izitaki ezijulile ziqeqesheka. Izinqamuleli zomazisi awengezi amapharamitha engeziwe; kuphela uma osayizi bokufaka nokukhishwayo behluka lapho ukuqagela okuncane (1x1 convolution) kulungisa ubukhulu ngaphambi kokwengeza.
I-Mastering Residual Networks
Amanethiwekhi Esalela (ResNets) amanethiwekhi ajulile e-neural angeza 'ukweqa ukuxhumana' okuvumela izendlalelo zifunde ukulungiswa okuncane esikhundleni sokuguqulwa okugcwele. Leli qhinga elilula lenze kwaba nokwenzeka ukuqeqesha amanethiwekhi amakhulu ezendlalelo ezijulile, okuvusa ukunemba kokubonwa kwesithombe. Amanethiwekhi Esalela awokugeleza kokusebenza kombono wekhompuyutha okuhumusha noma okukhiqiza imidiya ebonakalayo ukuze ihlaziywe, isebenze, futhi isungule. Ukuze wakhe ukuqonda okujulile, phatha ama-Residual Networks njengemodeli yokusebenza, hhayi isici esisodwa: chaza imiphumela efiselekayo, ucacise ukucabanga, futhi uhlukanise lokho isistimu engakwenza ngokwethembeka kulokho okusadinga ukwahlulela kochwepheshe.
Empeleni, amaqembu aqinile asebenzisa ukunemba kwebhalansi Yenethiwekhi Yezinsalela nezingokoqobo zokusebenza njengekhwalithi yedatha, ukuhluka kokukhanya, nokuvumelana kwamalebula. Babhala imibandela yempumelelo ecacile, ukuhlola okuqhathaniswa nedatha engokoqobo nokugeleza komsebenzi, futhi baphindaphinde ngokusekelwe kumaphethini okuhluleka aqashiwe esikhundleni sokuwina kwebhentshimakhi yesikhathi esisodwa. Yilapho ukuqonda kwethiyori kuguquka kube amandla ahlala njalo kuwo wonke umkhiqizo, inqubomgomo, kanye nokusebenza.
I-Visual AI ingakwazi ukuhlola, ukutholwa, nokumaka imisebenzi esikalini. Ngesikhathi esifanayo, amalungelo ezithombe kanye nemvume kungaba ubungozi bomthetho uma ukutholakala kungacacile. Indlela eqine kakhulu iwukuhlanganisa isivinini sokuhlola nesiyalo sokuphatha: qhuba abashayeli bezindiza, bamba ubufakazi, ushicilele amalogi ezinqumo, futhi ubuyekeze izivikelo ngokuqhubekayo njengoba imodeli yokuziphatha, okulindelwe ngabasebenzisi, kanye nezimfuneko zokulawula zishintsha.
I-Strategic Impact
I-Visual AI ingakwazi ukuhlola, ukutholwa, nokumaka imisebenzi esikalini.
I-Visual AI ingakwazi ukuhlola, ukutholwa, nokumaka imisebenzi esikalini. Ekusetshenzisweni kwekhwalithi ephezulu, lokhu kuhunyushwa emithethweni yokusebenza elinganisekayo, imingcele yobunikazi, nemikhuba yokubuyekeza ephindelelayo ukuze amaqembu akwazi ukukala ukuzethemba esikhundleni sokukala ukungaqondakali.
Amathimba aqanjiwe angakwazi ukulinganisa imiqondo ngokushesha ngezibuyekezo ezimbalwa ezenziwa mathupha.
Amathimba aqanjiwe angakwazi ukulinganisa imiqondo ngokushesha ngezibuyekezo ezimbalwa ezenziwa mathupha. Ekusetshenzisweni kwekhwalithi ephezulu, lokhu kuhunyushwa emithethweni yokusebenza elinganisekayo, imingcele yobunikazi, nemikhuba yokubuyekeza ephindelelayo ukuze amaqembu akwazi ukukala ukuzethemba esikhundleni sokukala ukungaqondakali.
Imisebenzi ingasebenzisa amasiginali wesithombe nawevidiyo obekunzima ukuwenza ngaphambilini.
Imisebenzi ingasebenzisa amasiginali wesithombe nawevidiyo obekunzima ukuwenza ngaphambilini. Ekusetshenzisweni kwekhwalithi ephezulu, lokhu kuhunyushwa emithethweni yokusebenza elinganisekayo, imingcele yobunikazi, nemikhuba yokubuyekeza ephindelelayo ukuze amaqembu akwazi ukukala ukuzethemba esikhundleni sokukala ukungaqondakali.
Ukuqaliswa Komhlaba Wangempela
I-ImageNet classification backbones (ResNet-50, ResNet-101) esetshenziswa njengezici eziqeqeshelwe kusengaphambili zokufunda ukudlulisa
Ukutholwa kwesimila nesilonda kuzithombe ze-radiology ne-pathology kusetshenziswa izishumeki ezisuselwe ku-ResNet
Ukutholwa kwento kanye nezibonelo zezinhlaka zesegimenti ezifana ne-Faster R-CNN kanye ne-Mask R-CNN esebenzisa i-backbones ye-ResNet
Amapayipi okubona ozishayelayo ahlukanisa abahamba ngezinyawo, izimoto, nezimpawu ezivela kumafreyimu ekhamera
Amaphethini Okusebenzisa
Residual Networks in practice
Ama-backbones wesigaba se-ImageNet (ResNet-50, ResNet-101) asetshenziswa njengezici eziqeqeshelwe kusengaphambili ukuze kufundwe ukudlulisa.
I-ImageNet Classification Backbones (ResNet-50, ResNet-101) esetshenziswa njengezikhiqizi zesici eziqeqeshwe kusengaphambili zokudlulisa ukufunda Amathimba ngokuvamile athola imiphumela engcono lapho echaza imingcele yekhwalithi ngaphambili, egcina indlela yokukhuphuka komuntu yamakesi asemaphethelweni, futhi alandelele kokubili izinzuzo zokukhiqiza nezindleko zamaphutha ngokuhamba kwesikhathi.
Residual Networks in practice
Ukutholwa kwesimila nesilonda kuzithombe ze-radiology ne-pathology kusetshenziswa izishumeki ezisuselwe ku-ResNet.
Ukutholwa kwesimila nezilonda ezithombeni ze-radiology ne-pathology kusetshenziswa izifaki khodi ezisekelwe ku-ResNet Amathimba ngokuvamile athola imiphumela engcono uma echaza izilinganiso zekhwalithi ngaphambili, agcina indlela yokukhuphuka komuntu yamakesi asemaphethelweni, futhi alandelele kokubili izinzuzo zokukhiqiza nezindleko zamaphutha ngokuhamba kwesikhathi.
Residual Networks in practice
Ukutholwa kwento kanye nezibonelo zezinhlaka zesegmentation ezifana ne-Faster R-CNN ne-Mask R-CNN esebenzisa ama-backbones e-ResNet.
Ukutholwa kwezinto kanye nezinhlaka zokuhlukaniswa kwezibonelo ezifana ne-Faster R-CNN kanye ne-Mask R-CNN esebenzisa i-ResNet backbones Teams ngokuvamile athola imiphumela engcono uma echaza imingcele yekhwalithi ngaphambili, egcina indlela yokukhuphuka komuntu yamakesi asemaphethelweni, futhi alandelele kokubili izinzuzo zokukhiqiza nezindleko zamaphutha ngokuhamba kwesikhathi.
Residual Networks in practice
Amapayipi okubona ozishayelayo ahlukanisa abahamba ngezinyawo, izimoto, nezimpawu ezivela kumafreyimu ekhamera.
Amapayipi okubona ukuzishayelela ahlukanisa abahamba ngezinyawo, izimoto, nezimpawu kumafreyimu ekhamera Amaqembu ngokuvamile athola imiphumela engcono uma echaza izinga eliphezulu ngaphambili, egcina indlela yokukhuphuka yabantu yamakesi asemaphethelweni, futhi elandelela kokubili izinzuzo zokukhiqiza nezindleko zamaphutha ngokuhamba kwesikhathi.
Izingozi & Guardrails
Amalungelo ezithombe kanye nemvume kungaba ubungozi bezomthetho uma ukuvela kungacacile.
Ukusebenza kwemodeli kungahluka kukho konke ukukhanya, izibalo zabantu, kanye nezindawo.
Okuhle okungelona iqiniso kungase kungabonakali ngaphandle uma izinga lokuzethemba liqashelwa.
Ukuqalisa Umhlahlandlela
Chaza indlela yokwamukela yokunemba, ukukhumbula, nezindleko zamaphutha.
Chaza indlela yokwamukela yokunemba, ukukhumbula, nezindleko zamaphutha. Phatha isinyathelo ngasinye njengesango lobufakazi: uma imibandela ingafinyelelwa, misa ukukhishwa, vala igebe, bese unweba ukusetshenziswa.
Hlola ngedatha efana nezimo zangempela zokukhiqiza.
Hlola ngedatha efana nezimo zangempela zokukhiqiza. Phatha isinyathelo ngasinye njengesango lobufakazi: uma imibandela ingafinyelelwa, misa ukukhishwa, vala igebe, bese unweba ukusetshenziswa.
Engeza isibuyekezo somuntu ukuze uthole ukuzethemba okuphansi noma izibikezelo zomthelela omkhulu.
Engeza isibuyekezo somuntu ukuze uthole ukuzethemba okuphansi noma izibikezelo zomthelela omkhulu. Phatha isinyathelo ngasinye njengesango lobufakazi: uma imibandela ingafinyelelwa, misa ukukhishwa, vala igebe, bese unweba ukusetshenziswa.
Landelela ukukhukhuleka kwemodeli bese uqinisekisa kabusha ngemva kwezinguquko zekhamera noma zesethi yedatha.
Landelela ukukhukhuleka kwemodeli bese uqinisekisa kabusha ngemva kwezinguquko zekhamera noma zesethi yedatha. Phatha isinyathelo ngasinye njengesango lobufakazi: uma imibandela ingafinyelelwa, misa ukukhishwa, vala igebe, bese unweba ukusetshenziswa.