Getting Started with HGFormer

This document provides a brief intro of the usage of HGFormer.

Please see Getting Started with Detectron2 for full usage.

Evaluation with Pre-trained Models

Download models.

Cityscapes -> ACDC

python demo/inference.py --config-file configs/cityscapes/hgformer_swin_tiny_bs16_20k.yaml \
--input datasets/acdc/rgb_anon/all/test --output path_to_output \
--opts MODEL.WEIGHTS path_to_checkpoint

After running the command, you will find the results on path_to_output. Then you can follow the instructions on ACDC evaluation server to get your scores. You can replace all with a specific type fog, snow, night, rain, if you want to evaluate on a specific type

Cityscapes -> Cityscapes-c

python test_city_c_level5.py --num-gpus 8 --config-file configs/city_c/hgformer_swin_tiny_bs16_20k.yaml \
 --eval-only MODEL.WEIGHTS path_to_checkpoint OUTPUT_DIR path_to_output

Cityscapes -> Others

python plain_train_net.py --num-gpus 8 --config-file configs/cityscapes/hgformer_swin_tiny_bs16_20k.yaml \
--eval-only MODEL.WEIGHTS path_to_checkpoint OUTPUT_DIR path_to_output

Mapillary -> Others

python plain_train_net.py --num-gpus 8 --config-file configs/mapillary/hgformer_swin_tiny_bs16_20k_mapillary.yaml \
--eval-only MODEL.WEIGHTS path_to_checkpoint OUTPUT_DIR path_to_output

Training in Command Line

To train a model, first setup the corresponding datasets following datasets/README.md, then prepare the models pre-trained on ImageNet classificaiton following tools/README.md. Finally run:

python plain_train_net.py --num-gpus 8 \
  --config-file configs/cityscapes/hgformer_swin_tiny_bs16_20k.yaml OUTPUT_DIR path_to_output

The configs are made for 8-GPU training. Since we use ADAMW optimizer, it is not clear how to scale learning rate with batch size. To train on 1 GPU, you need to figure out learning rate and batch size by yourself:

python plain_train_net.py \
  --config-file configs/cityscapes/hgformer_swin_tiny_bs16_20k.yaml \
  --num-gpus 1 SOLVER.IMS_PER_BATCH SET_TO_SOME_REASONABLE_VALUE SOLVER.BASE_LR SET_TO_SOME_REASONABLE_VALUE