Ishusho Yanditse
Kwandika amashusho nigikorwa cyo guhita utanga interuro-mvugo karemano isobanura ibiri mumashusho.
Incamake
It bridges vision and language, turning pixels into words that explain content, objects, and actions.
Kwibira cyane
Sisitemu yanditseho amashusho ifata ishusho hanyuma igatanga ibisobanuro neza nka 'imbwa yumukara ifata frisbee kumyatsi.' Sisitemu yo hambere yahujije urusobekerane rwakuyemo ibintu bigaragara hamwe numuyoboro usubiramo (LSTM) wabyaye amagambo icyarimwe, akenshi uyoborwa nubwitonzi kuburyo icyitegererezo 'gisa' mukarere kerekeranye na buri jambo. Sisitemu zigezweho zikoresha kodegisi ya transformateur iyerekwa hamwe na decoder ya transformateur y'ururimi, hamwe na moderi nini yerekana icyerekezo nka BLIP-2 na GPT-4V irashobora kwandika amashusho hamwe no kuvuga neza. Amahugurwa ashingiye ku mibare nka MS COCO, aho buri shusho ifite ibisobanuro byinshi byanditswe n'abantu. Ubwiza bupimirwa hamwe na metero nka CIDEr, BLEU, hamwe na CLIPScore ishingiye.
Ubushishozi
Abanditse byinshi bakurikiza kodegisi-decoder. Kodegisi ihindura ishusho mubice biranga ibice; decoder itanga amagambo atabishaka, guhanura buri kimenyetso giteganijwe kumashusho namagambo yatanzwe mbere. Icyitonderwa kireka decoder uburemere bwibice bitandukanye byamashusho kumagambo, kunoza ishingiro. Amahugurwa akoresha cross-entropy kumutwe wukuri-ukuri, rimwe na rimwe bigakurikirwa no gushimangira imyigishirize ihindura ibisobanuro byanditse-byiza nka CIDEr kugirango bigabanye kubogama.
Ingaruka z'Ingamba
Umuvuduko n'igipimo
AI igaragara irashobora gukora igenzura, gutahura, no gutondekanya imirimo kurwego.
Build choices
Amakipe arema arashobora prototype ibitekerezo byihuse hamwe nintoki nkeya.
Itsinda hamwe nakazi
Ibikorwa birashobora gukoresha amashusho nibimenyetso bya videwo byari bigoye gutunganya.
Igihe kizaza cyanditseho amashusho
Ibisobanuro biri guhuriza hamwe mubyerekezo rusange byururimi-rudasobanura gusa ahubwo rusubiza ibibazo, impamvu, no gukurikiza amabwiriza yerekeye amashusho. Tegereza ibintu byinshi, ibisobanuro byinshi bishobora kugenzurwa (uburebure bushobora guhinduka, imiterere, cyangwa intumbero), uburyo bwiza bufatika bwo gukumira ibintu bya salusine, hamwe nibikoresho bikomeye bigerwaho byerekana isi igaragara mugihe nyacyo. Indimi nyinshi na videwo byanditse bizaguka, kandi kuri moderi yibikoresho bizazana ibyigenga, ibisobanuro byihuse kuri terefone hamwe nudukoresho kubakoresha impumyi kandi batabona neza.
Gushyira mu bikorwa Isi
Gukora alt-text ibisobanuro byamafoto kugirango abasomyi ba ecran barashobora gufasha abakoresha impumyi kandi batabona neza
Kwiyerekana-byanditseho ibisobanuro hamwe nibishobora gushakishwa kumasomero manini yububiko bwamafoto hamwe nububiko-shusho
Gusobanura ibidukikije mu ijwi riranguruye ukoresheje porogaramu nka Microsoft Kubona AI cyangwa Kuba Amaso yanjye
Kwerekana amakadiri ya videwo hamwe nibisobanuro byanditse kugirango ushoboze gushakisha ibikubiyemo no kugereranya kurwego
Ingaruka & Kurinda
Uburenganzira bwishusho hamwe no kwemererwa birashobora guhinduka ibyago byemewe n'amategeko niba ibimenyetso bidasobanutse.
Imikorere yicyitegererezo irashobora gutandukana kumurika, demografiya, nibidukikije.
Ibyiza byibinyoma birashobora kutamenyekana keretse niba ibyiringiro byateganijwe bikurikiranwa.
Igishushanyo mbonera
Sobanura ibipimo byo kwemererwa kugiciro, kwibutsa, nibiciro byamakosa.
Gerageza hamwe namakuru ajyanye nuburyo nyabwo bwo gukora.
Ongeraho isubiramo ryabantu kubwizere buke cyangwa guhanura cyane.
Kurikirana icyitegererezo cya drift hanyuma uhindurwe nyuma ya kamera cyangwa dataset ihinduka.
Komeza Ubushakashatsi
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 Image Captioning 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
Ubuyobozi bukurikira
FLUX Ishusho Yerekana
Ibibazo bikunze kubazwa
What is Image Captioning?
Kwandika amashusho nigikorwa cyo guhita utanga interuro-mvugo karemano isobanura ibiri mumashusho. Ihuza icyerekezo nururimi, ihindura pigiseli mumagambo asobanura ibirimo, ibintu, nibikorwa.
Niki sisitemu yo kwandika amashusho itanga nkibisohoka?
Ishusho yerekana amashusho itanga interuro yinyandiko isobanura ibiri mumashusho.
Mubisanzwe byanditseho umuyoboro, ni uruhe ruhare kodegisi igaragara?
Kodegisi (akenshi CNN cyangwa iyerekwa rihindura) ihindura ishusho mubice biranga ururimi decoder ikoresha.
Ni ukubera iki kwitondera ari ingirakamaro mu kwandika amashusho?
Icyitonderwa gifasha decoder 'kureba' ibice byingenzi byishusho mugihe utanga buri jambo, kunoza ishingiro.
Nihe dataset ikoreshwa cyane mumahugurwa no gusuzuma amashusho yanditse?
MS COCO itanga amashusho buriwese hamwe ninyandiko nyinshi zanditswe n'abantu, bigatuma iba igipimo cyanditse.
Bisobanura iki kuri decoder yerekana ibisobanuro 'autoregressively'?
Igisekuru cya Autoregressive gitanga ibimenyetso icyarimwe, buri kintu cyateganijwe mbere yikimenyetso nishusho.