mmdetection/configs/rtmdet/classification

CSPNeXt ImageNet Pre-training

In this folder, we provide the imagenet pre-training config of RTMDet's backbone CSPNeXt.

Requirements

To train with these configs, please install MMClassification 1.x first.

Install by MIM:

mim install mmcls>=1.0.0rc0

or install by pip:

pip install mmcls>=1.0.0rc0

Prepare Dataset

To pre-train on ImageNet, you need to prepare the dataset first. Please refer to the guide.

How to Train

You can use the classification config in the same way as the detection config.

For single-GPU training, run:

python tools/train.py \
    ${CONFIG_FILE} \
    [optional arguments]

For multi-GPU training, run:

bash ./tools/dist_train.sh \
    ${CONFIG_FILE} \
    ${GPU_NUM} \
    [optional arguments]

More details can be found in user guides.

Results and Models

Model	resolution	Params(M)	Flops(G)	Top-1 (%)	Top-5 (%)	Download
CSPNeXt-tiny	224x224	2.73	0.34	69.44	89.45	model
CSPNeXt-s	224x224	4.89	0.66	74.41	92.23	model
CSPNeXt-m	224x224	13.05	1.93	79.27	94.79	model
CSPNeXt-l	224x224	27.16	4.19	81.30	95.62	model
CSPNeXt-x	224x224	48.85	7.76	82.10	95.69	model