English | 简体中文

Installation

Introduction

This document covers how to install PaddleDetection, its dependencies (including PaddlePaddle), together with COCO and Pascal VOC dataset.

For general information about PaddleDetection, please see README.md.

Install PaddlePaddle

Requirements:

  • Python2 or Python3 (Only support Python3 for windows)
  • CUDA >= 9.0
  • cuDNN >= 7.6
  • nccl >= 2.1.2

If you need GPU multi-card training, firstly please install NCCL. (Windows does not support nccl).

PaddleDetection depends on PaddlePaddle version relationship:

PaddleDetection version PaddlePaddle version tips
v0.3 >=1.7 --
v0.4 >= 1.8.4 PP-YOLO依赖1.8.4
v0.5 >= 1.8.4 Most models can run with >= 1.8.4, Cascade R-CNN and SOLOv2 depend on 2.0.0.rc

If you want install paddlepaddle, please follow the instructions in installation document.

Please make sure your PaddlePaddle installation was successful and the version of your PaddlePaddle is not lower than required. Verify with the following commands.

# To check PaddlePaddle installation in your Python interpreter
>>> import paddle.fluid as fluid
>>> fluid.install_check.run_check()

# To check PaddlePaddle version
python -c "import paddle; print(paddle.__version__)"

Other Dependencies

COCO-API:

COCO-API is needed for running. Installation is as follows:

git clone https://github.com/cocodataset/cocoapi.git
cd cocoapi/PythonAPI
# if cython is not installed
pip install Cython
# Install into global site-packages
make install
# Alternatively, if you do not have permissions or prefer
# not to install the COCO API into global site-packages
python setup.py install --user
# or with pip
pip install "git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI"

Installation of COCO-API in windows:

# if cython is not installed
pip install Cython
# Because the origin version of cocoapi does not support windows, another version is used which only supports Python3
pip install git+https://github.com/philferriere/cocoapi.git#subdirectory=PythonAPI

PaddleDetection

Clone Paddle models repository:

You can clone PaddleDetection with the following commands:

cd <path/to/clone/PaddleDetection>
git clone https://github.com/PaddlePaddle/PaddleDetection.git

Install Python dependencies:

Required python packages are specified in requirements.txt, and can be installed with:

pip install -r requirements.txt

Make sure the tests pass:

python ppdet/modeling/tests/test_architectures.py

Datasets

PaddleDetection includes support for COCO and Pascal VOC by default, please follow these instructions to set up the dataset.

Create symlinks for local datasets:

Default dataset path in config files is dataset/coco and dataset/voc, if the datasets are already available on disk, you can simply create symlinks to their directories:

ln -sf <path/to/coco> <path/to/paddle_detection>/dataset/coco
ln -sf <path/to/voc> <path/to/paddle_detection>/dataset/voc

For Pascal VOC dataset, you should create file list by:

python dataset/voc/create_list.py

Download datasets manually:

On the other hand, to download the datasets, run the following commands:

  • COCO
python dataset/coco/download_coco.py

COCO dataset with directory structures like this:

dataset/coco/
├── annotations
│   ├── instances_train2014.json
│   ├── instances_train2017.json
│   ├── instances_val2014.json
│   ├── instances_val2017.json
│   |   ...
├── train2017
│   ├── 000000000009.jpg
│   ├── 000000580008.jpg
│   |   ...
├── val2017
│   ├── 000000000139.jpg
│   ├── 000000000285.jpg
│   |   ...
|   ...
  • Pascal VOC
python dataset/voc/download_voc.py

Pascal VOC dataset with directory structure like this:

dataset/voc/
├── trainval.txt
├── test.txt
├── label_list.txt (optional)
├── VOCdevkit/VOC2007
│   ├── Annotations
│       ├── 001789.xml
│       |   ...
│   ├── JPEGImages
│       ├── 001789.jpg
│       |   ...
│   ├── ImageSets
│       |   ...
├── VOCdevkit/VOC2012
│   ├── Annotations
│       ├── 2011_003876.xml
│       |   ...
│   ├── JPEGImages
│       ├── 2011_003876.jpg
│       |   ...
│   ├── ImageSets
│       |   ...
|   ...

NOTE: If you set use_default_label=False in yaml configs, the label_list.txt of Pascal VOC dataset will be read, otherwise, label_list.txt is unnecessary and the default Pascal VOC label list which defined in voc_loader.py will be used.

Download datasets automatically:

If a training session is started but the dataset is not setup properly (e.g, not found in dataset/coco or dataset/voc), PaddleDetection can automatically download them from COCO-2017 and VOC2012, the decompressed datasets will be cached in ~/.cache/paddle/dataset/ and can be discovered automatically subsequently.

NOTE: