InoteveraGaidhi rinotevera
SwinIR Transformer Kudzoreredza
AI Yekutarisa
Gwara reVisual AI
DETR Transformer Kuonekwa
DETR (DEtection TRAnsformer) inogadziridza kutariswa kwechinhu seyakananga set-prediction dambudziko rinogadziriswa neshanduko, kubvisa nhanho dzakagadzirwa nemaoko senge mabhokisi eanchor uye isiri-yakanyanya kudzvinyirira.
Teerera kugwaro iriAI-yakarondedzerwa · ~2 min
Pa peji ino2 min verenga
Pfupiso
Izvo zvine basa nekuti zvakapa kuona kwakachena, kumagumo-kusvika-kumagumo pombi iyo yakafuridzira mafungu eshanduro-yakavakirwa-muono modhi.
Kudzika Kwakadzika
Yakaunzwa neFacebook AI muna 2020, DETR inosanganisa CNN musana neshanduro encoder-decoder. Iyo CNN inobvisa maficha emifananidzo; iyo encoder inosanganisa mamiriro epasi rose pamufananidzo wese; uye decoder inotora yakagadziriswa seti yezvakadzidzwa 'mivhunzo yechinhu' uye inoshandura imwe neimwe kuita chinhu chaonekwa (kirasi pamwe nebhokisi rinosungirirwa) kana 'hapana chinhu' mhedzisiro. Chinhu chakanyanya kukosha ndeye bipartite matching: panguva yekudzidziswa algorithm yeHungary inowana imwe-kune-imwe basa pakati pekufungidzira uye pasi-chokwadi zvinhu, saka modhi inodzidza kuburitsa rakasiyana bhokisi pachinhu chakananga. Izvi zvinobvisa kusanyanya kudzvinyirira uye anchor tuning. Kutengeserana kwacho kwaive kunonoka kuchinjika uye kushomeka kwechinhu chidiki, icho chinotevera senge Deformable DETR yakagadziriswa.
Kuitwa muNyika yeChokwadi
Kutsvaga uye tsiva vanofamba netsoka uye mota mune yakazvimirira-kutyaira yekutsvagisa dataset
Kusimbisa panoptic segmentation kana yawedzerwa kune-pixel mask kufanotaura
Kushanda semusana wezvivakwa zveakavhurika-mazwi uye zvigadziriso zvepasi
Kutsvaga zvinhu mumasherufu ezvitoro mapikicha pasina kugadzirisa saizi dzeanchor pane dataset
Ramangwana reDETR Transformer Detection
DETR yakatanga mhuri yese yekuona ma transformer. Misiyano yakadai seDeformable DETR, DAB-DETR, DN-DETR, uye DINO yakamhanyisa zvinoshamisa kudzidziswa uye nekunatsiridza huroyi, ine maDINO-maitiro anosvika pamusoro pemabhenji ekuonekwa. Iyo yemubvunzo-yakavakirwa, yekupedzisira-kusvika-kumagumo paradigm ikozvino inotambanukira kune segmentation, yekutevera, uye 3D yekuona, uye yakavhurika-mazwi ekutaridza anovaka pairi. Tarisira kuenderera mberi kwekusangana kwekuona, kupatsanura, uye kudzika kwemutauro kuva magadzirirwo emagetsi akabatana, DETR inorangarirwa sedanho rakakosha rakabvisa heuristics yakagadzirwa nemaoko.
Kuzvishandisa
Nei zvichikosha
- Kumhanya uye chiyero. Visual AI inogona kuita otomatiki yekuongorora, yekuona, uye yekumaka mabasa pachiyero.
- Vaka sarudzo. Zvikwata zvekugadzira zvinogona prototype pfungwa nekukurumidza nekudzokororwa kwemaoko mashoma.
- Team uye workflow. Mashandisirwo anogona kushandisa masaini emifananidzo nemavhidhiyo ayo aimbove akaoma kugadzirisa.
Chenjera
- Kodzero dzemifananidzo uye kubvumirwa kunogona kuve njodzi dzepamutemo kana provenance isina kujeka.
- Kuita kwemuenzaniso kunogona kusiyanisa kupenya, huwandu hwevanhu, uye nharaunda.
- Manyepo enhema anogona kusacherechedzwa kunze kwekunge zvikumbaridzo zvekuvimba zvikatariswa.
Matanho anotevera
- Tsanangura maitiro ekugamuchirwa echokwadi, kurangarira, uye mutengo wekukanganisa.
- Edzai nedata rinoenderana nemamiriro chaiwo ekugadzira.
- Wedzera ongororo yemunhu kune yakaderera-kusavimbika kana yakakwirira-inokanganisa kufanotaura.
- Tevera modhi kudonha uye simbisa mushure mekuchinja kwekamera kana dataset.
Ramba Uchiongorora
Tsamba yemahara
Tora iyo yemazuva ese AI pfupiso
Nhatu dzakasimbiswa dzeAI nyaya mangwanani ega ega pakati pevhiki, dzakanyorwa neChirungu chakajeka. Yemahara zvachose, hapana ads.
Imwe email zuva rega rega revhiki. Bvisa kunyoresa kamwe chete. Hatimbotengese kana kugovera kero yako.
Zviedze iwe pachako
Take the DETR Transformer Detection quiz
Mhinduro yepakarepo pamhinduro yega yega, uye chitupa chinogovaniswa chine ID inobatika kana wapedza kosi.
Tsigira dzidzo yemahara yeAI. AI Understanding ndeye 501(c)(3) isingabatsiri - hapana ads, hapana paywall, zvachose. Ita chipo
Mibvunzo inowanzo bvunzwa
Chii chinonzi DETR Transformer Detection?
DETR (DEtection TRAnsformer) inogadziridza kutariswa kwechinhu seyakananga set-prediction dambudziko rinogadziriswa neshanduko, kubvisa nhanho dzakagadzirwa nemaoko senge mabhokisi eanchor uye isiri-yakanyanya kudzvinyirira. Izvo zvine basa nekuti zvakapa kuona kwakachena, kumagumo-kusvika-kumagumo pombi iyo yakafuridzira mafungu eshanduro-yakavakirwa-muono modhi.
DETR inobvisa chii kana ichienzaniswa neyechinyakare ma detectors seFaster R-CNN?
DETR ndeyekupedzisira-kusvika-kumagumo uye inofanotaura seti yemabhokisi zvakananga, kubvisa anchors uye iyo isiri-yakanyanya kudzvinyirira post-processing nhanho.
Ndeipi algorithm inoshandiswa neDETR kuenzanisa fungidziro kune pasi-chokwadi zvinhu panguva yekudzidziswa?
DETR inoshandisa Hungarian algorithm kuumba imwe-kune-imwe assignment pakati pekufungidzira uye pasi-yechokwadi zvinhu zvekurasikirwa kwayo.
Chii chinonzi DETR's 'object queries'?
Mivhunzo yechinhu inhamba yakatarwa yevakadzidza vectors; iyo decoder inoshandura imwe neimwe kuita fungidziro yechinhu kana 'hapana chinhu' mhedzisiro.
Ndechipi chivakwa chese chinosanganiswa neDETR?
DETR pairs a convolutional musana wezvimiro zvine transformer encoder-decoder yekuseta kufanotaura.
How does DETR's training objective support inference without a separate NMS stage?
The learning objective encourages a unique prediction per object; this is a design and training mechanism rather than an infallibility guarantee.
Ramba uchidzidza
Mazano anoenderana
Mamwe madhairekitori akasarudzirwa nyaya iyi