Zurfin Komai Zurfin Monocular
DepthAnything samfurin tushe ne wanda ke kimanta nisa kowane pixel daga hoto na yau da kullun, ba tare da kayan masarufi na musamman ba.
Dubawa
It made robust, general-purpose depth sensing cheap and accessible for anything from phones to robots.
Zurfafa nutsewa
DepthAnything (2024, wanda masu bincike suka fitar ciki har da waɗanda ke TikTok/ByteDance da HKU) suna magance ƙima mai zurfi na monocular: tsinkayar taswira mai zurfi daga hoton RGB guda ɗaya. Ci gabansa ya kasance ma'auni: maimakon dogaro da iyakataccen bayanan zurfin da ake samu, ƙungiyar ta ƙera injin wanda ya yi wa kansa lakabi da hotuna kusan miliyan 62 da ba a lakabi ta hanyar amfani da ƙirar malami, sannan ta horar da ɗalibi kan wannan babbar ƙungiyar. Wannan yana ba da juzu'i mai ƙarfi na sifili a cikin gida, waje, da wuraren da ba a saba gani ba. Sakamakon asali na asali yana fitar da zurfin dangi (waɗanda pixels suke kusa ko nesa, ba daidaitattun mita ba). DepthAnything V2 (tsakiyar 2024) ya kaifafa cikakkun bayanai ta hanyar horar da malami kan bayanan roba tare da cikakkiyar gaskiya ta gaskiya, sannan karkatar da hotuna na gaske, gyara gefuna masu duhu da kurakuran abu.
Fahimtar Fasaha
Yana amfani da DINov2 mai canza hangen nesa-mai canza hangen nesa yana ciyar da babban tsinkayar irin salon DPT. Makullin dabara ita ce distillation da ake kula da shi: malami wanda ya horar da kan lakabin bayanan karya-lakabin miliyoyin hotuna marasa lakabi, kuma dalibi yana koyo daga duka biyun. V2 yana musanya haƙiƙanin alamun surutu don bayanan roba tare da cikakkiyar zurfin pixel, sannan ya mayar da baya zuwa ainihin hotuna, yana kawar da rashi da hayaniyar bayanan zurfin bayanai yayin kiyaye iyakoki.
Dabarun Tasiri
Gudu da sikelin
Kayayyakin AI na iya sarrafa aiki da bincike, ganowa, da ayyuka masu alama a sikelin.
Gina zaɓuɓɓuka
Ƙungiyoyin ƙirƙira za su iya samar da ra'ayoyi cikin sauri tare da ƙarancin bita da hannu.
Ƙungiya da aikin aiki
Ayyuka na iya amfani da siginar hoto da bidiyo waɗanda a baya suke da wahalar aiwatarwa.
Makomar Zurfin Komai Zurfin Halitta
Yi tsammanin haɗin kai cikin gilashin AR, kyamarori na wayowin komai da ruwan, da injiniyoyin mutum-mutumi inda LiDAR keɓe yana da tsada ko girma. Bambance-bambancen ma'auni waɗanda ke fitar da mita na gaskiya, da ƙirar bidiyo tare da kwanciyar hankali na ɗan lokaci (babu flicker tsakanin firam ɗin), suna ci gaba da sauri. Yayin da waɗannan samfuran ke raguwa don aiki akan na'urar a ainihin lokacin, tsinkayen kyamarar 3D guda ɗaya zai zama iyawar tsoho, ciyar da lissafin sararin samaniya, kewayawa mai cin gashin kansa, da sake fasalin yanayin 3D.
Aiwatar da Gaskiyar Duniya
Ƙirƙirar taswira mai zurfi don fitar da ɓacin rai na gaskiya (bokeh) a cikin hotunan wayar ruwan tabarau guda ɗaya.
Samar da hasashe na cikas na 3D ga jiragen sama marasa tsada da mutummutumi waɗanda ba su da LiDAR ko kyamarori na sitiriyo.
Ƙirƙirar taswirorin sanyaya zurfafa don ControlNet don haka masu samar da hoto su adana lissafi na yanayi.
Mayar da hotuna da fina-finai na 2D zuwa 3D ko tasirin parallax don VR da nunin stereoscopic.
Hatsari & Tsare-tsare
Haƙƙoƙin hoto da yarda na iya zama haxarin doka idan ba a fayyace ba.
Ayyukan samfuri na iya bambanta a ko'ina cikin haske, ƙididdiga, da mahalli.
Ƙarya tabbataccen ƙila ba za a iya lura da shi ba sai dai idan an kula da ƙofofin amincewa.
Taswirar Hanya
Ƙayyade ma'auni na karɓa don daidaito, tunowa, da farashi na kuskure.
Gwada tare da bayanan da suka dace da ainihin yanayin samarwa.
Ƙara bita na ɗan adam don ƙarancin amincewa ko tsinkaya mai tasiri.
Bi diddigin ƙirar ƙira kuma sake ingantawa bayan canje-canjen kamara ko saitin bayanai.
Ci gaba da Bincike
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the DepthAnything Monocular Depth quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Jagora na gaba
Ƙimar Zurfin Juna
Tambayoyin da ake yawan yi
What is DepthAnything Monocular Depth?
DepthAnything samfurin tushe ne wanda ke kimanta nisa kowane pixel daga hoto na yau da kullun, ba tare da kayan masarufi na musamman ba. Ya yi ƙarfi, zurfin maƙasudin maƙasudi mai arha da samun dama ga komai daga wayoyi zuwa mutummutumi.
Menene ma'anar zurfin kimanta 'monocular'?
Ma'anar monocular ana faɗi zurfin daga hoton kamara ɗaya (mono), ba tare da sitiriyo biyu ko firikwensin aiki ba.
Menene babban dabarar DepthAnything don cimma ƙarfi gama gari?
Tawagar ta kera injin bayanai wanda aka yi wa lakabi da dubun-dubatar hotuna marasa lakabi, yana fadada ma'aunin horo sosai.
Wanne encoder na baya DepthAnything yake ginawa akai?
DepthAnything yana amfani da DINov2 ViT encoder wanda aka haɗe tare da babban tsinkaya mai nau'in DPT.
Wane irin zurfin asalin DepthAnything ne da farko ke fitarwa?
Samfurin tushe yana tsinkayar yin odar zurfafa dangi maimakon cikakken nisan awo.
Ta yaya DepthAnything V2 ya inganta kyakkyawan daki-daki da gefuna?
V2 ya horar da malami akan hotunan roba tare da zurfin zurfin zurfi, sannan aka karkatar da su zuwa ainihin hotuna don iyakoki masu kaifi.