Reseau transformateur spatial
Reseau Transformateur Spatial (STNs) ay module yuñ mëna jàng lañu, yuy may reso neuronal bi mu deforme, wëlbatiku, dagg wala mu defaraat limu dugal ngir mëna bàyyi xelam ci li am solo.
Résumé
They give CNNs a built-in sense of spatial attention and invariance.
Plongeur bu xóot
Reseau convolutionnel yiñ miin duñu soppeeku ci coppite yi ci position, scale ak wëréelu, ñuy wéeru ci pooling ngir am tuuti muñ. Reseau transformatëri barab, Jaderberg ak ñeneen ñoo ko dugal. ci 2015, defar lii ci dugal benn modulu differensiaal buy def coppite geometrik bu leer ci kàrtu man-man yi. Module bi amna ñatti pàcc: ab reso lokalisasioŋ buy wax luy waaja am ci parametru coppite yi, ab generatëru griy buy tabax ab griy buy jël misaal ci parametre yooyu, ak ab sampler buy interpole liñu dugal ci poñ griy yi. Ndax jéego bu nekk mën nañu ko wuutale, transformatër bi yépp ñu ngi koy tàggat ci backpropagation te kenn duko saytu. Reseau bi dafay jàng, ci misaal, ni ñuy jubbanti ay lim yuñ tiim wala ñuy zoom ci gox biñ laaj, gëna dëgëral njub ak dëgër.
Gis-gis xarala
Reseau lokalisasioŋ bi dafay génne ay parametre (matrix affine 2x3) ngir tekki, eskaal, wëréelu ak cisage. Generatëru griy bi dafay delloosi pixel bu nekk ci coordonnée source bi jaaraleko ci matrix bi. Ginaaw loolu sampler bi dafay jàng liñu dugal ci interpolation bilineaire, muy luñu mëna wuutale suko defee gradient yi di jaar ci reso lokalisasioŋ bi. Loolu dafay may modle bi mu jàng coppite yi ci ñàkkum liggéey bi, di toppatoo ba noppi di kanonikalise gox yi ci laale.
njeextalu pexe
Gaawaay ak yaatuwaay
Visual IA mën na otomatise saytu, gis ak etiketu liggéey ci eskaal.
Tabax tànneef
Ekipu kreatif yi mën nañu defar konsept yu gëna gaaw te duñu def lu bari ci loxo.
Ekip ak def liggéey
Liggéeyukaay yi mën nañu jëfandikoo siñaal nataal wala wideo yu jafewoon lool ci liggéey.
Ëlëgu reso transformatëri barab
STNs indi nañu coppite ci ni reso yi di jëfandikoo geometrie ak bàyyi xel, di dundee ci convolution yuñ mëna deforme ak module yuñ jàngee deforme. Bi transformatër yiy bàyyi seen xel ci seen bopp ñoo ëpp doole, STN-style sampling différentielle mingi wéy ci liggéey yi soxla njubluwaay geometrik bu leer: xàmmee mbind, xaaj bu baax, ak normalisasioŋ pose. Xaarandil warping bu wuute ngir wéy di feeñ ci gis-gis 3D, rendu neuronal, ak bind nataalu medsin, lu bari boole ci bàyyi xel moo gën ñu koy wecci.
Doxal ci àdduna dëgg
Jubbanti ak jubbanti mbind mu xawa sëgg wala mu wëlbatiku balaa ñu koy xàmmee ci sistemu OCR mbindu seen
Zoom ci gox yu wuute (lu melni bakkan wala laaf picc) ngir mëna tànnal nataal bu baax
Normalise pose kanam ak jubluwaay ni jéego bu njëkk ci pipeline xàmmee kanam
Jubbanti jaar-jaar yi ak méngale scan yi ci bindu nataali pajum
Risk yi ak balustrade yi
Yelleefi nataal ak nangu mën na nekk risku yoon sudee fi ñu bawoo leerul.
Performance model bi mën na wuute ci leeraay bi, demographie bi ak environmaa bi.
Njuumteg positive yi mën nañu dem te kenn duko seetlu fileek xool wuñu buntu wóolu sa bopp.
Roadmap ngir samp gi
Mandargal kritërium nangug njub, woowaat ak njëgu njuumte.
Saytu ak done yu méngoo ak anam yi ñuy liggéeyee dëgg.
Yokk jàngat nit ngir xam fu wóorul dara wala am njeexital yu rëy.
Toppal model drift bi nga baaxal ko ginaaw bi kamera bi wala done yi soppeekoo.
Weyal di banneexu
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 Spatial Transformer Networks 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
Gis bi ci topp
gis-gis trafo DETR
Laaj yi ñuy faral di laaj
What is Spatial Transformer Networks?
Reseau Transformateur Spatial (STNs) ay module yuñ mëna jàng lañu, yuy may reso neuronal bi mu deforme, wëlbatiku, dagg wala mu defaraat limu dugal ngir mëna bàyyi xelam ci li am solo. Dañuy jox CNN yi yëg-yëgu bàyyi xel ci barab bi ak baña soppiku.
Ban kàttan la Reseau Transformateur Spatial di yokk ci Reseau neuronal bi?
STNs yi dañuy dugal benn modle buñu mëna jàng bu mëna tekki, wëlbati, escaleer, wala deforme kàrtu man-man yi ngir mëna bàyyi seen xel ci ëmbiit li am solo.
Yan ñatti mbir ñoo bokk ci benn modulu transformatër spasiaal?
STN amna reso buy lokalisasioŋ (di wax luy waaja am ci parametre yi), generatëru griy (defar coordonnées échantillonnage), ak échantillonnage (di interpoler liñ dugal).
Lan moo waral ñu mëna tàggat ab reso transformateur spatial ak backpropagation bu amul benn werante?
Interpolation bilineaire ci sampler bi mën nañu ko wuutale, moo tax gradient yi dañuy jaar ci generatëru griy bi dem ci reso lokalisasioŋ bi, loolu mooy tax ñu mëna tàggat seen bopp.
Lan la reso lokalisasioŋ bi di faral di génne ci coppite affine?
Ngir soppi affine, reso lokalisasioŋ bi dafay wax jiroom benn valeur yuñ raññe ni matrix 2x3 buy kode tekki, eskaal, wëréel, ak cisage.
Lan moo waral CNN yiñ miin duñu soppiku ci coppite yi am ci barab bi, di ñaax STN yi?
CNN yi duñu def lenn ludul invariance spatial bu ndaw jaaraleko ci pooling; STNs yokk anam wu leer, mën jàng ngir jëfandikoo position, scale, ak wëréelu.