Qoto dheer Wax kasta oo qoto dheer monocular
DepthAnything waa moodeel aasaasi ah oo qiyaasaya inta uu pixel kasta ka fog yahay hal sawir oo caadi ah, oo aan lahayn qalab gaar ah.
Dulmar
It made robust, general-purpose depth sensing cheap and accessible for anything from phones to robots.
quusid qoto dheer
DepthAnything (2024, oo ay sii daayeen cilmi-baarayaal ay ku jiraan kuwa TikTok/ByteDance iyo HKU) waxay wax ka qabtaan qiyaasta qoto dheer ee monocular: saadaalinta khariidadda qoto dheer ee hal sawir RGB. Horumarkeedu waxa uu ahaa miisaan: Halkii ay ku tiirsanaan lahaayeen oo keliya xogta qoto dheer ee la calaamadeeyay ee xaddidan, kooxdu waxay dhistay matoor si toos ah ugu calaamadiyay qiyaastii 62 milyan oo sawirro ah oo aan la calaamadin iyadoo la adeegsanayo moodal macallin, ka dibna arday u tababaray korpuskan weyn. Tani waxay siinaysaa guud ahaan eber-shot xoog badan guud ahaan gudaha, dibadda, iyo muuqaallada aan caadiga ahayn. Soo saarida asalka ah waxay soo saartaa qoto dheer (kuwaas oo pixels u dhow ama ka sii fog yahay, ma aha mitir sax ah). DepthAnything V2 (bartamihii 2024) waxay xoojisay tafaasiisha wanaagsan iyagoo macalinka ku tababaraya xogta synthetic oo leh run- run kaamil ah, ka dibna u jeexjeexa sawirada dhabta ah, hagaajinta cidhifyada qallafsan iyo khaladaadka shay ee hufan.
Aragtida Farsamada
Waxay adeegsataa koodiyaha aragga-beddelaha DINov2 oo quudinaysa madaxa saadaasha cufan ee qaabka DPT. Khiyaamada furaha ah waa kala-fogeynta hoose ee korjoogteynta: macalinka loo tababaray xogta calaamadeysan ee been abuurka ah wuxuu ku calaamadiyaa malaayiin sawirro ah oo aan calaamadeysan, ardayguna wuxuu wax ka bartaa labadaba. V2 waxay ku beddeshaa sumadaha dhabta ah ee buuqa badan xogta synthetic oo leh qoto dheer oo pixel ah, ka dibna waxay dib ugu celisaa sawirada dhabta ah, iyada oo ka leexanaysa yaraanta iyo sawaxanka tafsiiryada qoto dheer ee dhabta ah iyada oo la ilaalinayo xuduudaha qallafsan.
Saamaynta Istiraatijiyadeed
Xawaaraha iyo miisaanka
Visual AI wuxuu si otomaatig ah u samayn karaa baadhista, ogaanshaha, iyo sumadaynta hawlaha miisaanka.
Xulashada dhismayaasha
Kooxaha hal-abuurka leh waxay hindise karaan fikradaha si dhakhso leh iyagoo leh dib-u-eegis buugeed yar.
Kooxda iyo socodka shaqada
Hawlgalladu waxay isticmaali karaan calaamadaha muuqaalka iyo muuqaalka kuwaas oo markii hore adkeyd in la farsameeyo.
Mustaqbalka Qoto dheer Wax kasta oo qoto dheer monocular
Filo is dhexgalka adag ee muraayadaha AR, kamaradaha casriga ah, iyo robotics halkaas oo LiDAR u heellan uu yahay mid aad qaali u ah ama weyn. Kala duwanaanshiyaha mitirka ee soo saara mitirka runta ah, oo lagu daray moodooyinka fiidyowga ah ee leh qoto dheer oo ku meel gaar ah (aan ka dhex libdheynin fiilooyinka), ayaa si degdeg ah u socda. Maaddaama moodooyinkan ay u yaraanayaan inay ku shaqeeyaan aaladda waqtiga dhabta ah, aragtida 3D-kaliya ee hal-kaamera ah waxay noqon doontaa karti aan caadi ahayn, xisaabinta boosaska quudinta, navigation madax-bannaan, iyo dib-u-dhiska goobta 3D.
Dhaqangelinta Adduunka-dhabta ah
Soo saarista khariidado qoto dheer si loo kaxeeyo blur asalka dhabta ah (bokeh) ee sawirada sawirka casriga ah ee lens-ka ah.
Bixinta aragtida caqabada 3D ee diyaaradaha aan duuliyaha lahayn iyo robots-ka ee aan lahayn LiDAR ama kamaradaha stereo.
Abuuritaanka khariidadaha qaboojinta qoto dheer ee ControlNet si ay soosaarayaasha sawiradu u ilaaliyaan joomatari goobta.
U beddelashada sawirada 2D iyo filimada 3D ama saamaynta parallax ee VR iyo bandhigyada stereoscopic.
Khatarta & Dariiqyada Ilaalada
Xuquuqda sawirka iyo ogolaanshaha waxay noqon kartaa khataro sharci ah haddii caddayntu aanay caddayn.
Waxqabadka moodeelku wuu ku kala duwanaan karaa iftiinka, tirakoobka, iyo deegaanka.
Wanaagga beenta ah waxa laga yaabaa inaan la dareemin ilaa xadka kalsoonida aan la kormeerin.
Qorshe Hawleedka Dhaqangelinta
Qeex shuruudaha aqbalida ee saxnaanta, dib u celinta, iyo kharashyada khaladka.
Ku tijaabi xogta ku habboon xaaladaha wax soo saarka dhabta ah.
Ku dar dib u eegis bini'aadamka si aad u hesho kalsoonida hoose ama saameeynta sare.
Lasoco moodeel dhaqaaqa oo dib u cusboonaysii kamarada ama xogta kaydinta ka dib.
Sii wad Sahaminta
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
Hagaha xiga
Qiyaasta Qoto dheer ee Monocular
Su'aalaha soo noqnoqda
What is DepthAnything Monocular Depth?
DepthAnything waa moodeel aasaasi ah oo qiyaasaya inta uu pixel kasta ka fog yahay hal sawir oo caadi ah, oo aan lahayn qalab gaar ah. Waxay ka dhigtay mid adag, qoto dheer ujeedo-guud oo dareen raqiis ah oo la heli karo shay kasta laga bilaabo telefoonada ilaa robotyada.
Waa maxay macnaha qoto-dheer ee 'monocular'?
Monocular micnaheedu waa qoto dheer waxaa laga soo qaatay hal (mono) sawir kamarad ah, oo aan lahayn lammaane isteeriyo ama dareen firfircoon.
Maxay ahayd DepthAnything istaraatijiyaddeeda ugu weyn ee lagu gaarey xog-ururin xooggan?
Kooxdu waxay dhistay mishiin xog ah oo been abuur ah ku calaamadiyay tobanaan milyan oo sawiro ah oo aan la calaamadin, taasoo si wayn u balaadhisay miisaanka tababarka.
Waa maxay codeerka laf dhabarta uu DepthAnything ku dhiso?
DepthAnything waxay isticmaashaa DINov2 ViT encoder oo lagu lammaaniyay madaxa saadaasha cufan ee qaabka DPT.
Waa maxay nooca qoto dheer ee asalka ah ee DepthAnything ugu horreyntii?
Qaabka salku wuxuu saadaaliyaa dalbashada qoto dheer ee qaraabo ah halkii uu ka ahaan lahaa masaafo mitir ah oo dhammaystiran.
Sidee DepthAnything V2 u hagaajiyay tafaasiisha wanaagsan iyo geesaha?
V2 waxay macalinka ku tababartay sawirada synthetic si qoto dheer oo sax ah, ka dibna loo sawiray sawirada dhabta ah ee xuduudaha fiiqan.