Nkọwapụta onyonyo
Ntinye aha onyonyo bụ ọrụ nke imepụta ahịrịokwu asụsụ okike na-akọwa ihe dị na eserese na-akpaghị aka.
Nchịkọta
It bridges vision and language, turning pixels into words that explain content, objects, and actions.
Ime miri emi
Usoro ntinye aha onyonyo na-ese onyinyo wee wepụta nkọwa dị mma dị ka 'nkịta aja aja na-ejide frisbee na ahịhịa.' Sistemu mmalite jikọtara netwọk convolutional nke weputara njirimara a na-ahụ anya na netwọk na-emegharị ugboro ugboro (LSTM) nke na-emepụta okwu n'otu n'otu, na-eduzikarị site na nlebara anya ka ihe nlereanya ahụ 'na-ele' anya na mpaghara dị mkpa maka okwu ọ bụla. Sistemu ọgbara ọhụrụ na-eji ihe ngbanwe mgbanwe maka ọhụụ na ihe ngbanwe maka asụsụ, yana ụdị asụsụ ọhụụ dị ka BLIP-2 na GPT-4V nwere ike ịkọwa onyonyo n'ụzọ dị egwu. Ọzụzụ na-adabere na datasets dị ka MS COCO, ebe onyonyo ọ bụla nwere ọtụtụ edemede mmadụ dere. A na-eji metrik tụọ ogo dị ka CIDer, BLEU na CLIPScore dabere na ntinye.
Nghọta nka nka
Ọtụtụ ndị na-ede akwụkwọ akụkọ na-agbaso ụkpụrụ ngbanwe-concode. Ihe ngbanwe ahụ na-atụgharị ihe onyonyo a ka ọ bụrụ otu akụkụ nke vectors; decoder na-ewepụta okwu n'onwe ya, na-ebu amụma nke ọ bụla nwere ọnọdụ na onyonyo a yana okwu ndị ewepụtara na mbụ. Nlebara anya na-eme ka onye nrụpụta ahụ tụọ mpaghara onyonyo dị iche iche n'otu mkpụrụ okwu, na-emeziwanye ntọala ala. Ọzụzụ na-eji cross-entropy na nkọwa okwu ala-eziokwu, mgbe ụfọdụ na-esochi nkuzi nkwado nke na-ebuli metric isiokwu dị mma dị ka CIDer ozugbo iji belata nhụsianya mkpughe.
Mmetụta atụmatụ
Ọsọ na ọnụ ọgụgụ
Visual AI nwere ike megharịa nyocha, nchọpụta na mkpado ọrụ n'ọtụtụ.
Mee nhọrọ
Otu ndị na-emepụta ihe nwere ike imepụta echiche ngwa ngwa site na ngbanwe akwụkwọ ntuziaka ole na ole.
Team na usoro ọrụ
Ọrụ nwere ike iji onyonyo na akara vidiyo siri ike ịhazi.
Ọdịnihu nke nkọwa onyonyo
Nkọwa okwu na-ejikọta n'ụdị asụsụ ọhụụ nke na-akọwaghị naanị kamakwa na-aza ajụjụ, iche echiche na soro ntuziaka gbasara onyonyo. Na-atụ anya nkwubi okwu ndị a na-achịkwa nke ọma (ogologo, ụdị, ma ọ bụ nlekwasị anya), ntọala eziokwu ka mma iji kwụsị ihe ndị a na-ahụ anya, yana ngwa nnweta siri ike nke na-akọwa ụwa a na-ahụ anya ozugbo. Nkọwa okwu ọtụtụ asụsụ na vidiyo ga-agbasawanye, na ụdị ngwaọrụ ga-eweta nkọwa nzuzo, ngwa ngwa na ekwentị na wearables maka ndị kpuru ìsì na ndị na-ahụ ụzọ dị ala.
Mmejuputa n'ezie n'ụwa
Ịmepụta nkọwa alt-ederede nke foto ka ndị na-agụ enyo nwere ike inyere ndị kpuru ìsì na ndị na-ahụ ụzọ dị ala aka
Nkebi okwu na-atụ aro na akpaaka na mkpado enwere ike ịchọ maka nnukwu ọba akwụkwọ foto na nyiwe ihe onyonyo ngwaahịa
Na-akọwa gburugburu gburugburu site na ngwa dị ka Microsoft Ịhụ AI ma ọ bụ Bụrụ Anya m
Na-edepụta fremu vidiyo nwere nkọwa ederede iji mee ka ọchụchọ ọdịnaya nwee ike ime ya na imeru ihe n'ọ̀tụ̀tụ̀
Ihe ize ndụ & okporo ụzọ nche
Ikike onyonyo na nkwenye nwere ike bụrụ ihe egwu dị n'iwu ma ọ bụrụ na edoghị anya.
Ọrụ nlereanya nwere ike ịdịgasị iche n'ofe ọkụ, igwe mmadụ, na gburugburu.
Enwere ike ghara ịhụ ihe dị mma ma ọ bụrụ na enyochaghị oke ntụkwasị obi.
Map mmejuputa
Kọwaa ụkpụrụ nnabata maka nkenke, icheta, na ụgwọ njehie.
Nwalee na data dabara na ọnọdụ mmepụta n'ezie.
Tinye nyocha mmadụ maka obere obi ike ma ọ bụ amụma mmetụta dị elu.
Sochie ihe nlere anya wee megharịa ka emechara mgbanwe igwefoto ma ọ bụ dataset.
Nọgide na-eme nchọpụta
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
Ntuziaka na-esote
Ụdị onyonyo FLUX
Ajụjụ a na-ajụkarị
What is Image Captioning?
Ntinye aha onyonyo bụ ọrụ nke imepụta ahịrịokwu asụsụ okike na-akọwa ihe dị na eserese na-akpaghị aka. Ọ na-ejikọta ọhụụ na asụsụ, na-atụgharị pikselụ ka ọ bụrụ okwu na-akọwa ọdịnaya, ihe na omume.
Kedu ihe usoro ntinye aha onyonyo na-emepụta dị ka mmepụta?
Ndekọ onyonyo na-ewepụta ahịrịokwu ederede na-akọwa ọdịnaya dị na eserese.
N'ime pipeline ndepụta okwu ochie, kedu ọrụ ihe ntinye anya na-arụ?
Ihe ngbanwe ahụ (na-abụkarị CNN ma ọ bụ ihe ntụgharị ọhụụ) na-atụgharị onyonyo ka ọ bụrụ vectors atụmatụ nke onye ngbanwe asụsụ na-eji wee na-eji.
Kedu ihe kpatara nlebara anya ji baa uru na ntinye onyonyo?
Nlebara anya na-enyere onye decoder aka 'ileba anya' akụkụ kachasị mkpa nke onyonyo a mgbe ị na-emepụta okwu ọ bụla, na-emeziwanye ntọala.
Kedu ihe ndekọ data a na-ejikarị maka ọzụzụ na ịtụle ntinye aha onyonyo?
MS COCO na-enye onyogho nke ọ bụla jikọtara ya na ọtụtụ isiokwu mmadụ dere, na-eme ka ọ bụrụ akara nrịbama aha ọkọlọtọ.
Kedu ihe ọ pụtara maka decoder isiokwu iji mepụta okwu 'autoregressively'?
Ọgbọ na-emegharị onwe ya na-emepụta akara n'otu n'otu, nke ọ bụla kwadoro na akara ngosi na ihe oyiyi.