AimLoggerHook-in-MMCV

This is a walkthrough of using aim in mmcv.

aim github repo: https://github.com/aimhubio/aim#track

install docker

docker is required to launch aim UI

official tutorial : https://docs.docker.com/engine/install/ubuntu/

or you could just copy n paste the followings (should be alright

# delete old version, it's okay if noth happen (it just means you didn't install any of them
$ sudo apt-get remove docker docker-engine docker.io containerd runc
# set up the repo so that you could install & update docker in the repo in the future
$ sudo apt-get update
$ sudo apt-get install \
    apt-transport-https \
    ca-certificates \
    curl \
    gnupg \
    lsb-release
# add docker's offical GPG key
$ curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg
# stable repo
$ echo \
  "deb [arch=amd64 signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] https://download.docker.com/linux/ubuntu \
  $(lsb_release -cs) stable" | sudo tee /etc/apt/sources.list.d/docker.list > /dev/null

# install docker engine
$ sudo apt-get update
$ sudo apt-get install docker-ce docker-ce-cli containerd.io
# test whether you've installed it successfully
$ sudo docker run hello-world

run the following python code

import docker
client = docker.from_env()
client.ping()

if you see requests.exceptions.ConnectionError: ('Connection aborted.', PermissionError(13, 'Permission denied')) it's because docker daemon requires superuser privilege. Following commands will allow non-root users to have access as well.

$ sudo usermod -aG docker $USER
$ sudo systemctl restart docker
restart your shell

Ensure installation is completed by re-running the 3-lines python code.

AimLoggerHook

entermmcv/runner/hooks/logger/（depending on where you install mmcv, the following is my path

$ cd /home/sunyx30/anaconda3/lib/python3.8/site-packages/mmcv/runner/hooks/logger/
$ git clone https://github.com/Yuxuan-Sun/AimLoggerHook-in-MMDet.git
# copy aim.py to /hooks/logger
$ cp AimLoggerHook-in-MMDet/aim.py .

the 'hparam' is self-defined in aim.py, remember to modify it according to your model

remember to add AimLoggerHook in __init__.py

config

change your log_config in config file as sth like the following

log_config = dict(
    interval=50, 
    hooks=[
        dict(type='TextLoggerHook'),
        dict(type='AimLoggerHook')])

run aim

$ aim init
# train model as usual, the following is my command 
$ CUDA_VISIBLE_DEVICES=1,2,3 PORT=29501 tools/dist_train.sh configs/datatang/retinanet_repvgg_fpn.py 3
$ aim up
# remember to execute `aim init` and `aim up` in the same directory

then you should see sth like

┌--------------------------------------------------------┐
        Aim UI collects anonymous usage analytics.
                Read how to opt-out here:
   https://github.com/aimhubio/aim#anonymized-telemetry
└--------------------------------------------------------┘
Running Aim UI on repo `/home/sunyx30/datatang_hand_detection/.aim`
Open http://127.0.0.1:43800
Press Ctrl+C to exit

view aim UI locally

ssh -N -f -L localhost:8888:localhost:43800 remoteUser@remoteIP

after entering your server password, just enter http://localhost:8888 in you local browser. (You could change the port as you desire)

sometimes if port is already in use(it might cause blank UI page), just kill it and reconnect

lsof -ti:8888 | xargs kill -