yjh0410 5b6ebce4bb modify RTCDet-v2		2 tahun lalu
..
README.md	6f4f753562 keep training YOLOX-L	2 tahun lalu
build.py	2eb02320fc release RTCDet-v1 & rename RTMDet to RTCDet	2 tahun lalu
loss.py	8d590a1158 modify RTCDet-v2	2 tahun lalu
matcher.py	8d590a1158 modify RTCDet-v2	2 tahun lalu
rtcdet_v2.py	5754e59167 debug RTCDet-v2	2 tahun lalu
rtcdet_v2_backbone.py	5b6ebce4bb modify RTCDet-v2	2 tahun lalu
rtcdet_v2_basic.py	5b6ebce4bb modify RTCDet-v2	2 tahun lalu
rtcdet_v2_head.py	5b6ebce4bb modify RTCDet-v2	2 tahun lalu
rtcdet_v2_neck.py	2eb02320fc release RTCDet-v1 & rename RTMDet to RTCDet	2 tahun lalu
rtcdet_v2_pafpn.py	2eb02320fc release RTCDet-v1 & rename RTMDet to RTCDet	2 tahun lalu
rtcdet_v2_pred.py	2eb02320fc release RTCDet-v1 & rename RTMDet to RTCDet	2 tahun lalu

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.