yjh0410 d8ff8b610a use AlignedSimOTA & QFL for RTCDet-v2		2 år sedan
..
README.md	6f4f753562 keep training YOLOX-L	2 år sedan
build.py	2eb02320fc release RTCDet-v1 & rename RTMDet to RTCDet	2 år sedan
loss.py	d8ff8b610a use AlignedSimOTA & QFL for RTCDet-v2	2 år sedan
matcher.py	d8ff8b610a use AlignedSimOTA & QFL for RTCDet-v2	2 år sedan
rtcdet_v2.py	d8ff8b610a use AlignedSimOTA & QFL for RTCDet-v2	2 år sedan
rtcdet_v2_backbone.py	b83f8dd9a9 modify RTCDet-v2	2 år sedan
rtcdet_v2_basic.py	80fddd7b61 design ELANBlock for FPN	2 år sedan
rtcdet_v2_head.py	d8ff8b610a use AlignedSimOTA & QFL for RTCDet-v2	2 år sedan
rtcdet_v2_neck.py	ebddefb2f0 update	2 år sedan
rtcdet_v2_pafpn.py	80fddd7b61 design ELANBlock for FPN	2 år sedan
rtcdet_v2_pred.py	d8ff8b610a use AlignedSimOTA & QFL for RTCDet-v2	2 år sedan

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.