I-Optical Flow
Ukugeleza kokubona kulinganisela ukuthi iphikseli ngalinye lihamba kanjani phakathi kwamafreyimu evidiyo alandelanayo, okukhiqiza imephu eminyene yamavektha anyakazayo.
Uhlolojikelele
It is how machines perceive movement, speed, and direction in video.
I-Deep Dive
Ukugeleza kwe-Optical kwabela umcibisholo omncane onyakazayo kuwo wonke amaphikseli, okuchaza lapho okubonakala kuhamba khona ukusuka kolunye uhlaka kuye kolandelayo. Izindlela zakudala zincike ekucabangeni 'kokuqina kokuqina' - iphuzu ligcina ukukhanya okufanayo njengoba lihamba - kuhlanganiswe nezingqinamba zokushelela, njengaku-algorithms ye-Lucas-Kanade (sparse) kanye ne-Horn-Schunck (eminyene). Lezi zisebenza kahle ekunyakazeni okuncane, okuthambile kodwa zilwa nokunyakaza okusheshayo, ukuvaleka, nezindawo ezinkulu ezingathungwanga. Ukufunda okujulile kushintshe inkambu: amanethiwekhi afana ne-FlowNet, i-PWC-Net, futhi ikakhulukazi i-RAFT afunda ukufanisa izici kuwo wonke amafreyimu futhi acwengisise ngokuphindaphindiwe inkambu yokugeleza. Okuphumayo kushayela ukuqonda kwevidiyo noma kuphi lapho umbuzo ungekhona nje ukuthi 'yini ekuhlaka?' kodwa 'ihamba kanjani?'
I-Technical Insight
I-RAFT, indlela eyingqopha-mlando, yakha 'ivolumu yezindleko' ye-4D ebonisa ukuthi iphikseli ngalinye elikuhlaka olulodwa lifana kahle kangakanani nephikseli ngalinye kuhlaka lwesibili, bese isebenzisa isisebenzi sokubuyekeza esiphindelelayo (i-GRU) ukuze sicwenge isilinganiso sokugeleza ngezinyathelo eziningi ezincane - njengokuqhweba ngokuphindaphindiwe imicibisholo ebheke kumameshi angcono. Lokhu kuthuthukiswa okuphindaphindwayo, kunokuqagela okukodwa okukhulu, kunikeza ukugeleza okubukhali, okunembile ngisho nasekufudukeni okukhulu nemininingwane emihle, futhi kuhlanganisa kahle ezigcawini ezahlukene.
I-Strategic Impact
Isivinini nesikali
I-Visual AI ingakwazi ukuhlola, ukutholwa, nokumaka imisebenzi esikalini.
Yakha ukukhetha
Amathimba aqanjiwe angakwazi ukulinganisa imiqondo ngokushesha ngezibuyekezo ezimbalwa ezenziwa mathupha.
Ithimba kanye nokusebenza komsebenzi
Imisebenzi ingasebenzisa amasiginali wesithombe nawevidiyo obekunzima ukuwenza ngaphambilini.
Ikusasa Le-Optical Flow
Ukugeleza kwe-Optical kuya kusikhathi sangempela, ukulinganisa kokucaca okuphezulu kumadivayisi asemaphethelweni, ukuhlanganiswa okuqinile nokujula nokugeleza kwesigcawu se-3D, nokuqeqeshwa okuzigadile okufunda kuvidiyo eluhlaza ngaphandle kwamalebula abizayo eqiniso eliphansi. Njengoba amasistimu azimele namarobhothi efuna ukuqonda okucebile kokunyakaza, lindela ukuthi ukugeleza kuhlangane nokulandelelwa kwento nokubikezela ukuze imishini ingagcini nje ngokubona ukunyakaza kwamanje kodwa ilindele ukuthi izinto zizohamba kuphi ngokulandelayo, ngisho nangokuvaleka nokunyakaza kwamakhamera okusheshayo.
Ukuqaliswa Komhlaba Wangempela
Ukuzinzisa kwevidiyo kumafoni namakhamera esenzo akhansela ukunyakaza okubamba ngesandla okunyakazayo
Ukuhumusha kozimele okukhiqiza phakathi kozimele ukwenza ividiyo ibukeke ishelela noma isebenze ngokunensa
Izimoto ezisiza abashayeli kanye nezimoto ezizimele ezilinganisela isivinini nesiqondiso sezimoto eziseduze nabahamba ngezinyawo
Amakhodekhi okucindezela ividiyo abikezela ukunyakaza phakathi kozimele ukuze kugcinwe ividiyo kahle kakhulu
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.
Hlola ngedatha efana nezimo zangempela zokukhiqiza.
Engeza isibuyekezo somuntu ukuze uthole ukuzethemba okuphansi noma izibikezelo zomthelela omkhulu.
Landelela ukukhukhuleka kwemodeli bese uqinisekisa kabusha ngemva kwezinguquko zekhamera noma zesethi yedatha.
Qhubeka Uhlole
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 Optical Flow 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
Umhlahlandlela olandelayo
Ukufaniswa kokugeleza
Imibuzo evame ukubuzwa
What is Optical Flow?
Ukugeleza kokubona kulinganisela ukuthi iphikseli ngalinye lihamba kanjani phakathi kwamafreyimu evidiyo alandelanayo, okukhiqiza imephu eminyene yamavektha anyakazayo. Kuyindlela imishini ebona ngayo ukunyakaza, isivinini, kanye nesiqondiso kuvidiyo.
I-optical flow ilinganisela ini ngempela?
I-Optical flow ikhiqiza ama-motion vectors achaza ukuthi iphikseli ngalinye lihamba kanjani lisuka kolunye uzimele liye kolandelayo.
Kuyini ukucabangela 'kokuqina kokukhanya' okusetshenziswe ekugelezeni kokubona kwakudala?
Izindlela zakudala zithatha ukukhanya kwephoyinti elibonakalayo kuhlala kungashintshi njengoba lihamba, okubavumela ukuthi balifanise phakathi kozimele.
Iyiphi indlela yesimanjemanje yokufunda ngokujulile eyaziwa ngokulungiswa kokugeleza okuphindaphindayo kusetshenziswa umthamo wezindleko nezibuyekezo eziphindelelayo?
I-RAFT yakha umthamo wezindleko ze-4D futhi isebenzisa opharetha ophindayo ukuze icwenge ukugeleza ngezinyathelo eziningi ezincane, ifinyelele ukunemba kwesimanje.
Kungani izindlela zakudala zokugeleza kwe-optical zilwa nokunyakaza okukhulu, okusheshayo?
Izindlela ezakhelwe ekucabangeni kokunyakaza okuncane zilahlekelwa ukulandela lapho iphikseli igxuma kude phakathi kwamafreyimu, ibangele amaphutha.
Iluphi uhlelo lokusebenza oluncike ngokuqondile ekulinganiseni ukunyakaza phakathi kozimele?
Ukuhumusha kozimele kuhlanganisa ozimele abamaphakathi kusetshenziswa ukunyakaza kwephikseli okulinganiselwe, okukhiqiza ividiyo eshelelayo noma ehamba kancane.