Awesome
yolov6-sagemaker
This repo provides guide for training YOLOV6 with SageMaker.
Custom training container
Prerequisites
- Docker
- Aws cli
Steps to build and push custom container. For this example we are using yolov6s model.
- Install aws cli tool
# for debian based distros
sudo apt install awscli
Create the api access key from your AWS security credential dashboard.
Configure the aws account with the API access id created
aws configure
# paste the Key ID in the prompt
# paste the Access Key
# choose your region
# chosose json as output format
Login to access the DeepLearning Container image repository before pulling the image. Here are the list of <a href="https://github.com/aws/deep-learning-containers/blob/master/available_images.md">images</a>.
aws ecr get-login-password --region us-east-1 | docker login --username AWS --password-stdin 763104351884.dkr.ecr.us-east-1.amazonaws.com
#change the region as required
Pull the Base Pytorch image
docker pull 763104351884.dkr.ecr.us-east-1.amazonaws.com/pytorch-training:2.0.0-gpu-py310-cu118-ubuntu20.04-ec2
Clone the repo
git clone https://github.com/ashwincc/yolov6-sagemaker.git
cd yolov6-sagemaker/container
Login into your ecr
aws ecr get-login-password --region <region> | docker login --username AWS --password-stdin <account>.dkr.ecr.<region>.amazonaws.com
Build the docker image and tag.
REPO='yolov6s-sagemaker-training'
YOLO_IMAGE=$REPO:$(date '+%Y-%m-%d')
docker build . -t $YOLO_IMAGE
docker tag $YOLO_IMAGE <account>.dkr.ecr.<region>.amazonaws.com/$YOLO_IMAGE
Create the repository and push.
aws ecr create-repository --repository-name $REPO
docker push <account>.dkr.ecr.<region>.amazonaws.com/$YOLO_IMAGE
For using the custom container with sagemaker or testing it locally, please go through the main.ipynb notebook.
Todo:
- Add Tensorboard for training visualisation.