Segment Anything Model
Imodeli ye-Segment Anything (SAM) Meta iyimodeli eyisisekelo ye-AI yokuhlukaniswa kwesithombe: uma kunikezwe iphoyinti, ibhokisi, noma ukusikisela okuqinile, iveza ngokushesha into ehambisanayo.
Uhlolojikelele
It was built to generalize to objects and images it never saw during training, making segmentation a promptable task.
I-Deep Dive
Ikhishwe yi-Meta AI ngo-2023, i-SAM yenza kabusha ukuhlukaniswa njengenkinga esheshayo: uyinikeze ukwaziswa (ukuchofoza, ibhokisi, imaski, noma ukusikisela okususelwa embhalweni) futhi ibuyisela imaski yento eyodwa noma ngaphezulu. Amandla ayo avela ngokwengxenye esikalini: yaqeqeshwa ku-SA-1B, isethi yedatha yemaski engaphezu kwebhiliyoni ezithombeni eziyizigidi eziyi-11, eyakhiwe ngenjini yesichasiselo esiyimodeli-in-the-loop. Ngokwezakhiwo, i-SAM inesishumeki sesithombe esisindayo esisebenza kanye ngesithombe ngasinye, isifaki khodi esingasindi, kanye nesikhiphi sekhodi semaski esisheshayo, ngakho isithombe esisodwa esishumekiwe singaphinda sifundiswe ngokuhlanganyela ngesikhathi sangempela. Inika amandla ukudluliswa kwe-zero-shot emisebenzini eminingi. I-SAM 2, ekhishwe ngo-2024, inweba lokhu kuvidiyo, ilandelela izinto kuwo wonke amafreyimu.
I-Technical Insight
I-SAM isebenzisa isishumeki sesithombe se-Vision Transformer (ViT), ngokuvamile esiqeqeshelwa kusengaphambili ngombhalo wekhodi ozenzakalelayo omakiwe, ukuze ukhiqize ukushumeka kwesithombe esiminyene. Ukwaziswa kubhalwa ngekhodi kumathokheni, futhi idekhoda esekelwe ku-transformer ene-cross-attention fuses amathokheni okwaziswa ngesithombe esishumeka kumaski okukhiphayo kanye nezikolo zokuzethemba. Ukuze uxazulule ukungaqondakali (ukuchofoza kungasho inkinobho, ihembe, noma umuntu), i-SAM ibikezela imaski eminingana evumelekile ngesikhathi esisodwa futhi iwaklelise, ivumele ukusetshenziswa komfula noma ukwaziswa okwengeziwe kuhlukanise.
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 Lengxenye Noma Yini Imodeli
I-SAM isiphenduke umgogodla ozenzakalelayo wamathuluzi ezichasiselo, izithombe zezokwelapha, amarobhothi, namapayipi e-AR, ngokuvamile abhangqwa nezitholi noma amamodeli ombhalo wokugeleza komsebenzi wesilulumagama esivulekile 'sesegimenti ngegama'. Lindela okuhlukile okulula, okusheshayo (i-MobileSAM, i-EfficientSAM) ukuze isetshenziswe kudivayisi, ukuhlanganiswa okujulile nolimi ukuze kube nesegimenti eshayelwa umbhalo ngokugcwele, nokunwetshwa okuqhubekayo kuvidiyo ne-3D. Njengemodeli yesisekelo, ukushumeka kwayo kuya ngokuya kusetshenziswa kabusha njengesendlalelo sombono sondla amanye amasistimu.
Ukuqaliswa Komhlaba Wangempela
Izinkundla zesichasiselo sesithombe zisebenzisa i-SAM ukuvumela abalebula ukuthi bachofoze kanye futhi bazenzele ngokuzenzakalela imaski yento enembayo, ukusika isikhathi sokulebula.
Abacwaningi bavumelanisa i-SAM (isb., i-MedSAM) ukuze ichaze izitho nezimila kumaskeni we-CT kanye ne-MRI.
Abahleli bezithombe namavidiyo bahlanganisa i-SAM ukuze isike izihloko noma isuse ingemuva ngokuchofoza okukodwa.
Amathrekhi e-SAM 2 namasegimenti ezinto kuwo wonke amafreyimu wevidiyo wemiphumela ye-AR kanye nombono wamarobhothi.
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 Segment Anything Model 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
Amamodeli Avumelanayo
Imibuzo evame ukubuzwa
What is Segment Anything Model?
Imodeli ye-Segment Anything (SAM) Meta iyimodeli eyisisekelo ye-AI yokuhlukaniswa kwesithombe: uma kunikezwe iphoyinti, ibhokisi, noma ukusikisela okuqinile, iveza ngokushesha into ehambisanayo. Yakhelwe ukuhlanganisa izinto nezithombe engakaze ibone ngesikhathi sokuqeqeshwa, okwenza ukuhlukaniswa kube umsebenzi osheshayo.
Imuphi umbono obalulekile owenza i-SAM 'imodeli yesisekelo' yokuhlukaniswa?
I-SAM yenza kabusha isegimenti njengokuyalelwa futhi yayidizayinelwe ukudluliswa okungasho lutho ezintweni nezithombe ezingaphandle kwesethi yayo yokuqeqeshwa.
Ilinganiselwa ukuthi inkulu kangakanani idathasethi ye-SA-1B esetshenziselwa ukuqeqesha i-SAM?
I-SA-1B iqukethe imaski engaphezu kwebhiliyoni e-1 ezithombeni ezingaba yizigidi ezingu-11, ezakhiwe ngenjini yesichasiselo esiyimodeli-in-the-loop.
Kungani i-SAM ihlukanisa ukwakheka kwayo kusishumeki sesithombe esisindayo kanye nesishumeki/isikhiphi khodi esisheshayo?
Umbhalo wekhodi wesithombe obizayo usebenza kanye; bese ukubhala ngekhodi okusheshayo okushibhile kanye nokukhishwa kwekhodi kwemaski kuvumela ukuphinda kufakwe ngokushesha, okusebenzisanayo kwesithombe esifanayo.
I-SAM ikusingatha kanjani ukwaziswa okungacacile, njengokuchofoza okukodwa okungasho izinto ezimbalwa?
I-SAM ikhipha amamaski ekhandidethi amaningana anamaphuzu ukuze ukungaqondakali kuxazululwe ngezinga noma ukwaziswa okwengeziwe.
Iluphi uhlobo lomgogodla olusetshenziswa i-SAM ukufaka ikhodi yesithombe sokufaka?
Isishumeki sesithombe se-SAM siyi-Vision Transformer, evamise ukuqeqeshwa kusengaphambili ngombhalo wekhodi ozenzakalelayo omakiwe, okhiqiza ukushumeka kwesithombe esiminyene.