- https://github.com/open-mmlab/mmdetection # mmdet
- https://github.com/open-mmlab/mmcv
- https://github.com/facebookresearch/detectron2
- https://github.com/facebookresearch/fvcore
- https://github.com/ultralytics/yolov5
- https://github.com/fastai/fastai
- https://github.com/yitu-opensource/MobileNeXt
- https://github.com/pytorch/pytorch
- https://github.com/pytorch/vision
- https://github.com/torchgan/torchgan
- https://github.com/Lyken17/pytorch-OpCounter # thop(打印网络 flops与参数量)
- https://github.com/albumentations-team/albumentations
- https://github.com/aleju/imgaug
- https://github.com/bethgelab/imagecorruptions
# 计算 mAP
pip install pycocotools
tool for CV
!pip install mmcv-full==1.3.11 -f https://download.openmmlab.com/mmcv/dist/cu110/torch1.7.0/index.html
!pip install openmim > /dev/null
!mim install mmdet
!pip install fvcore>=0.1
!pip install pytorch-lightning>=1.3.8
!pip install timm>=0.4.12
!pip install pycocotools
!pip install pycuda
!pip install thop # 计算网络的flops
# apex 安装
git clone https://github.com/NVIDIA/apex.git
cd apex
pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./
# pip install -v --no-cache-dir --global-option="--pyprof" --global-option="--cpp_ext" --global-option="--cuda_ext" ./
# python setup.py install --cpp_ext --cuda_ext
!pip install https://codechina.csdn.net/wc781708249/toolcv/archive/master.zip
# or
!pip install 'git+https://codechina.csdn.net/wc781708249/toolcv.git'
!pip install 'git+https://codechina.csdn.net/wc781708249/toolcv.git@master' # @切换分支
# or
!python setup.py install
!pip install 'git+https://codechina.csdn.net/wc781708249/toolcv.git' > /dev/null
!pip install mmcv-full==1.3.11 -f https://download.openmmlab.com/mmcv/dist/cu110/torch1.7.0/index.html > /dev/null
!pip install openmim > /dev/null
!mim install mmdet > /dev/null
# pip install wandb
# Flexible integration for any Python script
import wandb
# 1. Start a W&B run
wandb.init(project='gpt3')
# 2. Save model inputs and hyperparameters
config = wandb.config
config.learning_rate = 0.01
# Model training here
# 3. Log metrics over time to visualize performance
wandb.log({"loss": loss})