yjh0410 ebddefb2f0 update		il y a 2 ans
..
README.md	6f4f753562 keep training YOLOX-L	il y a 2 ans
build.py	2eb02320fc release RTCDet-v1 & rename RTMDet to RTCDet	il y a 2 ans
loss.py	fca944e333 update	il y a 2 ans
matcher.py	8d590a1158 modify RTCDet-v2	il y a 2 ans
rtcdet_v2.py	5754e59167 debug RTCDet-v2	il y a 2 ans
rtcdet_v2_backbone.py	ebddefb2f0 update	il y a 2 ans
rtcdet_v2_basic.py	ebddefb2f0 update	il y a 2 ans
rtcdet_v2_head.py	fd8e9e5e0c update ELANNet-v2	il y a 2 ans
rtcdet_v2_neck.py	ebddefb2f0 update	il y a 2 ans
rtcdet_v2_pafpn.py	ebddefb2f0 update	il y a 2 ans
rtcdet_v2_pred.py	767ac95ea3 add aux loss for RTCDet-v2	il y a 2 ans

RTCDet-v2: My Second Empirical Study of Real-Time Convolutional Object Detectors.

For training, we train my RTCDetv2 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 RTCDetv2-X with the setting of batch size=128.