Computing Rectangular Haar Features for Cascade Detection Training

7/22/2019 Computing Rectangular Haar Features for Cascade Detection Training

1/8

B`dputegi[abtfgiuhfr Jffr

Lafturas l`rBfsbfoaOatabte`g _rfegegi

]e65?=?

Cuhy :, ;=62

7/22/2019 Computing Rectangular Haar Features for Cascade Detection Training

2/8

B`gtagts

B`gtagts

=.6 Egtr`oubte`g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

=.; Jffr Hena Lafturas . . . . . . . . . . . . . . . . . . . . . . . . .

=.2 Egtairfh Edfia [aprasagtfte`g . . . . . . . . . . . . . . . . . .

=.? Edphadagtfte`g Oatfehs . . . . . . . . . . . . . . . . . . . . . . .

=.1 B`oa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

[alaragbas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7/22/2019 Computing Rectangular Haar Features for Cascade Detection Training

3/8

B`dputegi [abtfgiuhfr Jffr Lafturas l`r Bfsbfoa Oatabte`g _rfegegi

B`dputegi [abtfgiuhfr JffrLafturas l`r Bfsbfoa Oatabte`g

_rfegegi

=.6 Egtr`oubte`g

Eg tja frtebha wa wehh h``n ft Jffr laftura axtrfbte`g wjebj es usaol`r bfsbfoa `kcabt oatabte`g. Fgo pr`veoa f iagareb datj`o `llaftura axtrfbte`g .

=.; Jffr Hena LafturasJffr hena Lafturas fra sedpha rabtfgiuhfr lafturas wjebj fra kfsao`g Jffr Yfvahats. _ja sat `l rabtfgiuhfr lafturas usao l`r vesufhrab`igete`g tfsn fra g`t fbutfhhy jffr wfvahat kut kafr s`da ra-sadkhagba t` Jffr wfvahat kfses lugbte`g.

_ja Jffr lafturas fra avfhuftao ft fhh p`ssekha trfgshfte`g fgooyofbteb sbfhas. _jus f vary hfria gudkar `l lafturas fra ax-trfbtao lr`d edfias.

B`gseoar tja l`hh`wegi Jffr lafturas. Leiura 6f rasadkhas f lerst

(f) Jffr Laftura 6 (k) Jffr Laftura ; (b) Jffr Laftura 2

`roar oarevfteva fh`gi tja x oerabte`g.Fhh p`ssekha trfgshfte`g fgooyofbteb sbfhas gaao t` ka b`gseoarao

7/22/2019 Computing Rectangular Haar Features for Cascade Detection Training

4/8

B`dputegi [abtfgiuhfr Jffr Lafturas l`r Bfsbfoa Oatabte`g _rfegegi

_ja gudkar trfgshfte`gs fh`gi y oerabte`g ft subbasseva sbfhas fra(;?,;2,. . . ,2,;,6) ,wjebj es t`tfh `l 2==.

_ja gudkar `l trfgshfte`gs fh`gi tja x oerabte`g ft subbassevasbfhas fra (;2,;6,65,. . . ,1,2,6),wjebj es t`tfh `l 6??

_ja t`tfh gudkar `l lafturas fra ?2;==.

Leiura 6k rasadkhas lerst `roar oarevfteva fh`gi tja y oerabte`g._ja bfhbuhfta `l lafturas es sfda fs Jffr laftura 6 fgo t`tfhgudkar `l lafturas fra ?2;==.

Leiura 6b rasadkhas sab`go `roar oarevfteva fh`gi tja x oerab-

te`g. B`gseoaregi fhh tja p`ssekha trfgshfte`g fgo sbfhas ._ja gudkar trfgshfte`gs fh`gi y oerabte`g ft subbasseva sbfhas fra(;?,;2,. . . ,2,;,6) ,wjebj es t`tfh `l 2==.

_ja gudkar `l trfgshfte`gs fh`gi tja x oerabte`g ft subbassevasbfhas fra (;;,65,6

7/22/2019 Computing Rectangular Haar Features for Cascade Detection Training

5/8

B`dputegi [abtfgiuhfr Jffr Lafturas l`r Bfsbfoa Oatabte`g _rfegegi

_ja gudkar `l trfgshfte`gs fh`gi tja x oerabte`g ft subbassevasbfhas fra (;2,;6,. . . ,1,2,6),wjebj es t`tfh `l 6??.

_ja t`tfh gudkar `l lafturas fra ;=92

7/22/2019 Computing Rectangular Haar Features for Cascade Detection Training

6/8

B`dputegi [abtfgiuhfr Jffr Lafturas l`r Bfsbfoa Oatabte`g _rfegegi

`l ; rabtfgiuhfr frafs. _ja pexahs eg khfbn fraf fra waeijtao ky-6 wjeha pexahs eg tja wjeta fraf fra waeijtao ky 6._ja fed es t`b`dputa tja sud `var fhh tja pexahs `l Jffr lafturas. _jes bfg ka

o`ga ky b`dputegi tja sud usegi egtairfh edfia `var tja khfbn fgowjeta rabtfgiuhfr fraf tjad sedphy waeijegi tja rasuhts ky +6 `r -6fgo fooegi et t` iat tja b`dphata sud.

_ja kfseb staps fra 3-

B`dputa sud `l pexahs l`r fhh rabtfgihas tjft dfna up JffrLaftura

Yaeijt tja rasuhts `l egtairfh sud b`dputfte`g `l egoeveoufhrabtfgihas

Foo tja rasuhts `l fhh egoeveoufh rabtfgihas t` iat tja legfhrasuht.

=.? Edphadagtfte`g Oatfehs

_ja

Bhfss es f kfseb uget. Et es bjfrfbtarezao ky rabtfgihafgo waeijt .

_ja

bhfss b`gsests `l vabt`r `l

Bhfssas.

_` b`dputa tja Jffr laftura h``p `var tja vabt`r `r

`kcabts b`dputa tja egtairfh sud,duhtephy ky tja rabtfgiha waeijt,fgobfhbuhfta tja buduhfteva sud `var fhh tja rabtfgihas.

_ja

bhfss b`gsests `l vabt`r `l `kcabts `l typa

.

Ievag f egput edfia wa b`dputa tja egtairfh edfia raprasagtf-te`g `l tja edfia.L`r _jes wa usa tja kueht-eg @pagB\ lugbte`g.

_ja

datj`o eg

bhfss fbbapts fs egputf edfia fgo b`dputa tja jfrr rabtfgiuhfr lafturas `var tja edfia.

_` degedeza tja allabt `l heijtegi b`goete`gs fhh tja sukwego`wswara vfrefgba g`rdfhezao.

\frefgba `l f sukwego`w bfg ka quenhy b`dputao usegi

7/22/2019 Computing Rectangular Haar Features for Cascade Detection Training

7/8

B`dputegi [abtfgiuhfr Jffr Lafturas l`r Bfsbfoa Oatabte`g _rfegegi

wjara x es tja pexah wetjeg tja sukwego`w,d es tja dafg fgo estja stfgofro oavefte`g.

_jus wa gaao t` b`dputa tja dafg fgo squfrao sud `var tjasuk wego`w `var tja edfia.

Hena tja sud ,squfrao sud fhs` bfg ka b`dputao allebeagthy us-egi egtairfh edfia raprasagtfte`g.

_` b`dputa dafg fgo squfrao sud wa b`dputa tja sud `l pex-ahs `var tja sukwego`w l`r egtairfh edfias fgo squfra egtairfh edfia.

_ja wa foo tja rasuhts `var fhh tja suk rabtfgihas eg tja Jffr

laftura t` b`dputa tja dafg fgo squfra sud `l pexahs.Qsegi tjasa rasuhts wa bfg b`dputa tja vfrefgba `var tja suk-wego`ws.

_ja rasuht `l waeijtao sud `l pexahs es oeveoao ky tja vfrefgbat` parl`rd vfrefgba g`rdfhezfte`g.

_ja teda tfna t` b`dputa fhh tja jffr lafturas l`r f ;1x;1edfia es =.2 sab.

=.1 B`oa_ja b`oa l`r tja tastegi fgo trfegegi utehety bfg ka l`ugo ft

7/22/2019 Computing Rectangular Haar Features for Cascade Detection Training

8/8

Kekhe`irfpjy

Kekhe`irfpjy