Home Verkennen Help
Inloggen
junxiaoyao
/
RT-ODLab
1
0
Vork 0
Bestanden Issues 0 Pull-aanvragen 0 Wiki
Boom: 01b4ef8444
Aftakkingen Labels
RT-ODLab
/
models
/
detectors
/
rtcdet
yjh0410 01b4ef8444 debug AlignedSimOTA 2 jaren geleden
..
README.md ebe1d9e27f train RTCDet-S on COCO 2 jaren geleden
build.py 578dc295f4 redesign RTCDet 2 jaren geleden
loss.py 01b4ef8444 debug AlignedSimOTA 2 jaren geleden
matcher.py 222a4fadd1 debug AlignedSimOTA for RTCDet-M 2 jaren geleden
rtcdet.py f6753b0e9b train RTCDet-S 2 jaren geleden
rtcdet_backbone.py 4e429e1847 train RTCDet-S on COCO 2 jaren geleden
rtcdet_basic.py ae03c10fc1 train RTCDet-M with 320-960 2 jaren geleden
rtcdet_head.py 578dc295f4 redesign RTCDet 2 jaren geleden
rtcdet_neck.py 578dc295f4 redesign RTCDet 2 jaren geleden
rtcdet_pafpn.py 62a5b79e6a update 2 jaren geleden
rtcdet_pred.py f6753b0e9b train RTCDet-S 2 jaren geleden

README.md

RTCDet: My Empirical Study of Real-Time Convolutional Object Detectors.

Model Scale Batch APtest
0.5:0.95 		APtest
0.5 		APval
0.5:0.95 		APval
0.5 		FLOPs
(G) 		Params
(M) 		Weight
RTCDet-N 640 8xb16
RTCDet-T 640 8xb16
RTCDet-S 640 8xb16 30.9 8.5
RTCDet-M 640 8xb16 48.7 67.6 80.3 22.6 ckpt
RTCDet-L 640 8xb16
RTCDet-X 640 8xb16
  • For training, we train my RTCDet series series with 300 epochs on COCO.
  • For data augmentation, we use the large scale jitter (LSJ), Mosaic augmentation and Mixup augmentation, following the setting of YOLOX, but we remove the rotation transformation which is used in YOLOX's strong augmentation.
  • For optimizer, we use AdamW with weight decay 0.05 and base per image lr 0.001 / 64.
  • For learning rate scheduler, we use linear decay scheduler.
  • Due to my limited computing resources, I can not train RTCDet-X with the setting of batch size=128.
Model Scale Batch APval
0.5:0.95 		APval
0.5 		FLOPs
(G) 		Params
(M) 		Weight
RTCDet-P 320 8xb16 -
RTCDet-P 416 8xb16 -
RTCDet-P 512 8xb16 -
RTCDet-P 640 8xb16 -