Merge pull request #1 from modelai/ymir-dev

first commit for ymir

.dockerignore

@@ -0,0 +1,226 @@
+# Repo-specific DockerIgnore -------------------------------------------------------------------------------------------
+#.git
+.cache
+.idea
+runs
+output
+coco
+storage.googleapis.com
+
+data/samples/*
+**/results*.csv
+*.jpg
+
+ymir/tensorrt/build
+ymir/tensorrt/pt_result
+ymir/tensorrt/trt_result
+# Neural Network weights -----------------------------------------------------------------------------------------------
+#**/*.pt
+#**/*.pth
+**/*.pkl
+**/*.onnx
+**/*.engine
+**/*.mlmodel
+**/*.torchscript
+**/*.torchscript.pt
+**/*.tflite
+**/*.h5
+**/*.pb
+*_saved_model/
+*_web_model/
+*_openvino_model/
+
+# Below Copied From .gitignore -----------------------------------------------------------------------------------------
+# Below Copied From .gitignore -----------------------------------------------------------------------------------------
+
+
+# GitHub Python GitIgnore ----------------------------------------------------------------------------------------------
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+env/
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+wandb/
+.installed.cfg
+*.egg
+
+# PyInstaller
+# Usually these files are written by a python script from a template
+# before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+.hypothesis/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# pyenv
+.python-version
+
+# celery beat schedule file
+celerybeat-schedule
+
+# SageMath parsed files
+*.sage.py
+
+# dotenv
+.env
+
+# virtualenv
+.venv*
+venv*/
+ENV*/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+
+
+# https://github.com/github/gitignore/blob/master/Global/macOS.gitignore -----------------------------------------------
+
+# General
+.DS_Store
+.AppleDouble
+.LSOverride
+
+# Icon must end with two \r
+Icon
+Icon?
+
+# Thumbnails
+._*
+
+# Files that might appear in the root of a volume
+.DocumentRevisions-V100
+.fseventsd
+.Spotlight-V100
+.TemporaryItems
+.Trashes
+.VolumeIcon.icns
+.com.apple.timemachine.donotpresent
+
+# Directories potentially created on remote AFP share
+.AppleDB
+.AppleDesktop
+Network Trash Folder
+Temporary Items
+.apdisk
+
+
+# https://github.com/github/gitignore/blob/master/Global/JetBrains.gitignore
+# Covers JetBrains IDEs: IntelliJ, RubyMine, PhpStorm, AppCode, PyCharm, CLion, Android Studio and WebStorm
+# Reference: https://intellij-support.jetbrains.com/hc/en-us/articles/206544839
+
+# User-specific stuff:
+.idea/*
+.idea/**/workspace.xml
+.idea/**/tasks.xml
+.idea/dictionaries
+.html # Bokeh Plots
+.pg # TensorFlow Frozen Graphs
+.avi # videos
+
+# Sensitive or high-churn files:
+.idea/**/dataSources/
+.idea/**/dataSources.ids
+.idea/**/dataSources.local.xml
+.idea/**/sqlDataSources.xml
+.idea/**/dynamic.xml
+.idea/**/uiDesigner.xml
+
+# Gradle:
+.idea/**/gradle.xml
+.idea/**/libraries
+
+# CMake
+cmake-build-debug/
+cmake-build-release/
+
+# Mongo Explorer plugin:
+.idea/**/mongoSettings.xml
+
+## File-based project format:
+*.iws
+
+## Plugin-specific files:
+
+# IntelliJ
+out/
+
+# mpeltonen/sbt-idea plugin
+.idea_modules/
+
+# JIRA plugin
+atlassian-ide-plugin.xml
+
+# Cursive Clojure plugin
+.idea/replstate.xml
+
+# Crashlytics plugin (for Android Studio and IntelliJ)
+com_crashlytics_export_strings.xml
+crashlytics.properties
+crashlytics-build.properties
+fabric.properties

mmyolo/engine/hooks/__init__.py

@@ -1,10 +1,11 @@
 # Copyright (c) OpenMMLab. All rights reserved.
 from .ppyoloe_param_scheduler_hook import PPYOLOEParamSchedulerHook
 from .switch_to_deploy_hook import SwitchToDeployHook
+from .ymir_training_monitor_hook import YmirTrainingMonitorHook
 from .yolov5_param_scheduler_hook import YOLOv5ParamSchedulerHook
 from .yolox_mode_switch_hook import YOLOXModeSwitchHook
 
 __all__ = [
  'YOLOv5ParamSchedulerHook', 'YOLOXModeSwitchHook', 'SwitchToDeployHook',
- 'PPYOLOEParamSchedulerHook'
+ 'PPYOLOEParamSchedulerHook', 'YmirTrainingMonitorHook'
 ]

mmyolo/engine/hooks/ymir_training_monitor_hook.py

@@ -0,0 +1,77 @@
+"""
+hook for ymir training process, write the monitor.txt, save the latest model
+"""
+import glob
+import logging
+import os.path as osp
+import re
+import warnings
+from typing import Dict, Optional, Union
+
+from mmengine.hooks import Hook
+from mmengine.registry import HOOKS
+from ymir_exc.util import (get_merged_config, write_ymir_monitor_process,
+ write_ymir_training_result)
+
+
+@HOOKS.register_module()
+class YmirTrainingMonitorHook(Hook):
+ """
+ for epoch based training loop only.
+
+ 1. write monitor.txt
+ 2. save the latest checkpoint with id=last if exist, note the checkpoint maybe clear late.
+ 3. save the latest best checkpoint with id=best if exist, note the checkpoint maybe clear late.
+ """
+ # the priority should lower than CheckpointHook (priority = VERY_LOW)
+ priority = 'LOWEST'
+
+ def __init__(self, interval: int = 10):
+ self.interval = interval
+ self.ymir_cfg = get_merged_config()
+
+ def after_train_iter(self,
+ runner,
+ batch_idx: int,
+ data_batch: Optional[Union[dict, tuple, list]] = None,
+ outputs: Optional[dict] = None) -> None:
+ if runner.rank in [0, -1] and self.every_n_inner_iters(batch_idx, self.interval):
+ percent = (runner.epoch + batch_idx / len(runner.train_dataloader)) / runner.max_epochs
+ write_ymir_monitor_process(self.ymir_cfg, task='training', naive_stage_percent=percent, stage='task')
+
+ def after_val_epoch(self, runner, metrics: Optional[Dict[str, float]] = None) -> None:
+ """
+ metrics: {'coco/bbox_mAP': 0.001, 'coco/bbox_mAP_50': 0.003, 'coco/bbox_mAP_75': 0.0, 'coco/bbox_mAP_s': 0.0, 'coco/bbox_mAP_m': 0.0, 'coco/bbox_mAP_l': 0.001}
+
+ evaluation_result: {'mAP': 0.001, 'mAP_50': 0.003, ...}
+ """
+ if runner.rank in [0, -1]:
+ N = len('coco/bbox_')
+ evaluation_result = {key[N:]: value for key, value in metrics.items()} # type: ignore
+ out_dir = self.ymir_cfg.ymir.output.models_dir
+ cfg_files = glob.glob(osp.join(out_dir, '*.py'))
+
+ best_ckpts = glob.glob(osp.join(out_dir, 'best_coco', '*.pth'))
+ if len(best_ckpts) > 0:
+ newest_best_ckpt = max(best_ckpts, key=osp.getctime)
+ best_epoch = int(re.findall(r'\d+', newest_best_ckpt)[0])
+ # if current checkpoint is the newest checkpoint, keep it
+ if best_epoch == runner.epoch:
+ logging.info(f'epoch={runner.epoch}, save {newest_best_ckpt} to result.yaml')
+ write_ymir_training_result(self.ymir_cfg,
+ files=[newest_best_ckpt] + cfg_files,
+ id='best',
+ evaluation_result=evaluation_result)
+ else:
+ warnings.warn(f'no best checkpoint found on {runner.epoch}')
+
+ latest_ckpts = glob.glob(osp.join(out_dir, '*.pth'))
+ if len(latest_ckpts) > 0:
+ last_ckpt = max(latest_ckpts, key=osp.getctime)
+ logging.info(f'epoch={runner.epoch}, save {last_ckpt} to result.yaml')
+ write_ymir_training_result(self.ymir_cfg,
+ files=[last_ckpt] + cfg_files,
+ id='last',
+ evaluation_result=evaluation_result)
+ else:
+ warnings.warn(f'no latest checkpoint found on {runner.epoch}')

requirements.txt

@@ -1,3 +1,5 @@
 -r requirements/build.txt
 -r requirements/runtime.txt
--r requirements/tests.txt
+# -r requirements/tests.txt
+-r requirements/albu.txt
+-r requirements/ymir.txt

requirements/ymir.txt

@@ -0,0 +1,7 @@
+opencv-contrib-python>=4.0
+easydict
+tqdm
+imagesize
+nptyping
+tensorboard
+-e git+https://github.com/modelai/ymir-executor-sdk.git@ymir2.1.0#egg=ymir-exc

tools/train.py

@@ -7,9 +7,11 @@
 from mmengine.config import Config, DictAction
 from mmengine.logging import print_log
 from mmengine.runner import Runner
+from ymir_exc.util import get_merged_config
 
 from mmyolo.registry import RUNNERS
 from mmyolo.utils import register_all_modules
+from ymir.utils.common import modify_mmengine_config
 
 
 def parse_args():
@@ -61,6 +63,8 @@ def main():
 
  # load config
  cfg = Config.fromfile(args.config)
+ ymir_cfg = get_merged_config()
+ modify_mmengine_config(cfg, ymir_cfg)
  # replace the ${key} with the value of cfg.key
  # cfg = replace_cfg_vals(cfg)
  cfg.launcher = args.launcher

ymir/Dockerfile

@@ -0,0 +1,37 @@
+FROM pytorch/pytorch:1.11.0-cuda11.3-cudnn8-runtime
+
+# To fix GPG key error when running apt-get update
+RUN apt-get update && apt-get install -y gnupg2
+RUN apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/3bf863cc.pub
+RUN apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64/7fa2af80.pub
+
+# (Optional)
+RUN sed -i 's/http:\/\/archive.ubuntu.com\/ubuntu\//http:\/\/mirrors.aliyun.com\/ubuntu\//g' /etc/apt/sources.list && \
+ pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
+
+RUN apt-get update \
+ && apt-get install -y gcc ffmpeg libsm6 libxext6 git ninja-build libglib2.0-0 libxrender-dev vim \
+ && apt-get clean \
+ && rm -rf /var/lib/apt/lists/*
+
+RUN pip install openmim \
+ && mim install "mmengine>=0.3.1" \
+ && mim install "mmcv>=2.0.0rc1,<2.1.0" \
+ && mim install "mmdet>=3.0.0rc5,<3.1.0"
+
+# Install MMYOLO
+COPY . /app
+RUN cd /app && \
+ pip install --no-cache-dir -r requirements.txt && \
+ mkdir -p /img-man && \
+ mv ymir/img-man/*.yaml /img-man && \
+ mkdir /weights && \
+ mv ymir/weights/*.pth /weights && \
+ mkdir -p /root/.cache/torch/hub/checkpoints && \
+ mv ymir/weights/imagenet/*.pth /root/.cache/torch/hub/checkpoints
+
+ENV PYTHONPATH=.
+WORKDIR /app
+
+RUN echo "python3 ymir/start.py" > /usr/bin/start.sh
+CMD bash /usr/bin/start.sh