OEM New Common Rail Valve Gungano F00VC01329 Ye 0445110168 169 284 315 jekiseni

Kuonekwa kwekukanganisa kwechigaro chemotokari injector valve zvichibva pane fusion fusion(Chikamu 3)

Nekuda kweizvozvo, mukuonekwa kwechigaro chejekiseni chevharuvhu, mufananidzo unofanirwa kudzvanywa, uye saizi yemufananidzo inogadziriswa kusvika 800 × 600, mushure mekuwana iyo yakabatana yakajairwa data data, nzira yekusimudzira data inoshandiswa kudzivirira kushomeka kwedata. uye iyo modhi generalization kugona inowedzerwa.Kukwidziridzwa kwedata chikamu chakakosha chekudzidzisa mamodheru ekudzidza kwakadzama [3].Pane kazhinji nzira mbiri dzekuwedzera data.Imwe ndeyekuwedzera data perturbation layer kune network modhi kubvumira kuti chifananidzo chidzidziswe nguva dzese, pane imwe nzira yakatwasuka uye yakapusa, masampuli emifananidzo anosimudzirwa nekugadziriswa kwemifananidzo isati yadzidziswa, isu tinowedzera iyo data seti tichishandisa. nzira dzekusimudzira mufananidzo senge geometry nenzvimbo yeruvara, uye shandisa HSV munzvimbo yeruvara, sezvakaratidzwa mumufananidzo 1.

Kuvandudzwa kweFaster R-CNN defect defection model MuFaster R-CNN algorithm model, kutanga pane zvose, unofanira kubvisa zvinhu zvemufananidzo wekuisa, uye izvo zvakabviswa zvinobuda zvinogona kukanganisa zvakananga maitiro ekupedzisira ekuona.Mwoyo wekuonekwa kwechinhu kubviswa kwezvinhu.Iyo yakajairika ficha yekubvisa network muFaster R-CNN algorithm modhi ndiyo VGG-16 network.Iyi yetiweki modhi yakatanga kushandiswa muchikamu chemufananidzo [4], uyezve yakave yakanakisa musemantic segmentation [5] uye saliency yekuona [6].

Iyo ficha yekuwedzera network muFaster R-CNN algorithm modhi yakaiswa kuVGG-16, kunyangwe iyo algorithm modhi iine kuita kwakanaka pakuona, inongoshandisa iyo mepu inobuda kubva kune yekupedzisira layer mumifananidzo yekubvisa, saka kuchave kumwe kurasikirwa uye mepu yemaficha haigone kupedzwa zvizere, izvo zvinozotungamira mukusarurama mukuonekwa kwezvinhu zvidiki zvakanangwa uye kukanganisa yekupedzisira yekuzivikanwa maitiro.