Skip to content

Latest commit

 

History

History
32 lines (24 loc) · 729 Bytes

README.md

File metadata and controls

32 lines (24 loc) · 729 Bytes
Raw

Introduction

Codes for HuBMAP Challenge based on mmsegmentation

Usage

  • build docker image
cd docker
docker build -t mmsegmentation .
  • edit tools/docker_train.sh file

Repalce [DIR_TO_DATA] and [DIR_TO_REPO].

GPUID=$1
CONFIG=$2

docker run -it --gpus all --ipc=host \
  -v [DIR_TO_DATA]:/data/ \
  -v [DIR_TO_REPO]:/workspace/mmsegmentation/ \
  --rm mmsegmentation:latest \
  /bin/bash -c \
  "cd /workspace/mmsegmentation; pip install -v -e .; CUDA_VISIBLE_DEVICES=${GPUID} python tools/train.py ${CONFIG}"
  • train
sh tools/docker_train.sh [GPU_ID] [CONFIG_FILE]