Merge remote-tracking branch 'origin/master' into fixBug

.gitignore

@@ -112,6 +112,7 @@ data
 *.pkl.json
 *.log.json
 work_dirs/
+mmcls/.mim
 
 # Pytorch
 *.pth

MANIFEST.in

@@ -1,3 +1,3 @@
-include mmcls/model-index.yml
-recursive-include mmcls/configs *.py *.yml
-recursive-include mmcls/tools *.sh *.py
+include mmcls/.mim/model-index.yml
+recursive-include mmcls/.mim/configs *.py *.yml
+recursive-include mmcls/.mim/tools *.py *.sh

README.md

@@ -85,6 +85,7 @@ We wish that the toolbox and benchmark could serve the growing research communit
 ## Projects in OpenMMLab
 
 - [MMCV](https://github.com/open-mmlab/mmcv): OpenMMLab foundational library for computer vision.
+- [MIM](https://github.com/open-mmlab/mim): MIM Installs OpenMMLab Packages.
 - [MMClassification](https://github.com/open-mmlab/mmclassification): OpenMMLab image classification toolbox and benchmark.
 - [MMDetection](https://github.com/open-mmlab/mmdetection): OpenMMLab detection toolbox and benchmark.
 - [MMDetection3D](https://github.com/open-mmlab/mmdetection3d): OpenMMLab's next-generation platform for general 3D object detection.

README_zh-CN.md

@@ -72,6 +72,7 @@ MMClassification 是一款由不同学校和公司共同贡献的开源项目。
 ## OpenMMLab 的其他项目
 
 - [MMCV](https://github.com/open-mmlab/mmcv): OpenMMLab 计算机视觉基础库
+- [MIM](https://github.com/open-mmlab/mim): MIM 是 OpenMMlab 项目、算法、模型的统一入口
 - [MMDetection](https://github.com/open-mmlab/mmdetection): OpenMMLab 检测工具箱与测试基准
 - [MMDetection3D](https://github.com/open-mmlab/mmdetection3d): OpenMMLab 新一代通用 3D 目标检测平台
 - [MMSegmentation](https://github.com/open-mmlab/mmsegmentation): OpenMMLab 语义分割工具箱与测试基准

configs/resnet/resnet50_b64x32_warmup_label_smooth_imagenet.py

@@ -1,4 +1,4 @@
-_base_ = ['./resnet50_batch2048_warmup.py']
+_base_ = ['./resnet50_b64x32_warmup_imagenet.py']
 model = dict(
     head=dict(
         type='LinearClsHead',

docker/Dockerfile

@@ -0,0 +1,22 @@
+ARG PYTORCH="1.6.0"
+ARG CUDA="10.1"
+ARG CUDNN="7"
+
+FROM pytorch/pytorch:${PYTORCH}-cuda${CUDA}-cudnn${CUDNN}-devel
+
+ENV TORCH_CUDA_ARCH_LIST="6.0 6.1 7.0+PTX"
+ENV TORCH_NVCC_FLAGS="-Xfatbin -compress-all"
+ENV CMAKE_PREFIX_PATH="(dirname(which conda))/../"
+
+RUN apt-get update && apt-get install -y ffmpeg libsm6 libxext6 git ninja-build libglib2.0-0 libsm6 libxrender-dev libxext6 \
+    && apt-get clean \
+    && rm -rf /var/lib/apt/lists/*
+
+# Install MMCV
+RUN pip install mmcv-full==1.3.8 -f https://download.openmmlab.com/mmcv/dist/cu101/torch1.6.0/index.html
+
+# Install MMClassification
+RUN conda clean --all
+RUN git clone https://github.com/open-mmlab/mmclassification.git
+WORKDIR ./mmclassification
+RUN pip install --no-cache-dir -e .

docs/install.md

@@ -103,6 +103,23 @@ Note:
 
 you can install it before installing [mmcv](https://github.com/open-mmlab/mmcv).
 
+#### Another option: Docker Image
+
+We provide a [Dockerfile](/docker/Dockerfile) to build an image.
+
+```shell
+# build an image with PyTorch 1.6.0, CUDA 10.1, CUDNN 7.
+docker build -f ./docker/Dockerfile --rm -t mmcls:torch1.6.0-cuda10.1-cudnn7 .
+```
+
+**Important:** Make sure you've installed the [nvidia-container-toolkit](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html#docker).
+
+Run a container built from mmcls image with command:
+
+```shell
+docker run --gpus all --shm-size=8g -it -v {DATA_DIR}:/workspace/mmclassification/data mmcls:torch1.6.0-cuda10.1-cudnn7 /bin/bash
+```
+
 ### Using multiple MMClassification versions
 
 The train and test scripts already modify the `PYTHONPATH` to ensure the script use the MMClassification in the current directory.

docs_zh-CN/install.md

@@ -98,6 +98,23 @@ pip install -e .  # 或者 "python setup.py develop"
 
 2. 如果希望使用 `opencv-python-headless` 而不是 `opencv-python`，可以在安装 [mmcv](https://github.com/open-mmlab/mmcv) 之前提前安装。
 
+#### 利用 Docker 镜像安装 MMClassification
+
+MMClassification 提供 [Dockerfile](/docker/Dockerfile) ，可以通过以下命令创建 docker 镜像。
+
+```shell
+# 创建基于 PyTorch 1.6.0, CUDA 10.1, CUDNN 7 的镜像。
+docker build -f ./docker/Dockerfile --rm -t mmcls:torch1.6.0-cuda10.1-cudnn7 .
+```
+
+**注意:** 确保已经安装了 [nvidia-container-toolkit](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html#docker).
+
+运行一个基于上述镜像的容器:
+
+```shell
+docker run --gpus all --shm-size=8g -it -v {DATA_DIR}:/workspace/mmclassification/data mmcls:torch1.6.0-cuda10.1-cudnn7 /bin/bash
+```
+
 ### 在多个 MMClassification 版本下进行开发
 
 MMClassification 的训练和测试脚本已经修改了 `PYTHONPATH` 变量，以确保其能够运行当前目录下的 MMClassification。