Deformable 可变形的DETR

This repository is an official implementation of the paper Deformable DETR: Deformable Transformers for End-to-End Object Detection.

该存储库是论文《可变形DETR：用于端到端对象检测的可变形变压器》的正式实现。

https://github.com/fundamentalvision/deformable-detr

Introduction

Deformable DETR is an efficient and fast-converging end-to-end object detector. It mitigates the high complexity and slow convergence issues of DETR via a novel sampling-based efficient attention mechanism.

可变形DETR是一种高效且快速收敛的端到端对象检测器。通过一种新颖的基于采样的有效注意力机制，缓解了DETR的高复杂性和缓慢收敛的问题。

Abstract摘要

DETR has been recently proposed to eliminate the need for many hand-designed components in object detection while demonstrating good performance. However, it suffers from slow convergence and limited feature spatial resolution, due to the limitation of Transformer attention modules in processing image feature maps. To mitigate these issues, we proposed Deformable DETR, whose attention modules only attend to a small set of key sampling points around a reference. Deformable DETR can achieve better performance than DETR (especially on small objects) with 10× less training epochs. Extensive experiments on the COCO benchmark demonstrate the effectiveness of our approach.

最近提出了DETR，以消除目标检测中对许多手工设计组件的需求，同时表现出良好的性能。但是，由于Transformer注意模块在处理图像特征图时的局限性，它收敛缓慢且特征空间分辨率有限。为了缓解这些问题，提出了可变形DETR，其关注模块仅关注参考周围的一小部分关键采样点。可变形的DETR可以比DETR（尤其是在小物体上）获得更好的性能，训练时间减少10倍。在COCO Benchmark数据集上进行的大量实验证明了方法的有效性。

License

This project is released under the Apache 2.0 license.

项目是根据Apache 2.0许可发布的。

Changelog

See changelog.md for detailed logs of major changes.

有关主要更改的详细日志，请参见changelog.md。

Citing 引用可变形Deformable DETR

If you find Deformable DETR useful in your research, please consider citing:

如果发现Deformable可变形DETR在研究中很有用，考虑引用以下内容：

@article{zhu2020deformable,

title={Deformable DETR: Deformable Transformers for End-to-End Object Detection},

author={Zhu, Xizhou and Su, Weijie and Lu, Lewei and Li, Bin and Wang, Xiaogang and Dai, Jifeng},

journal={arXiv preprint arXiv:2010.04159},

year={2020}

}

Main Results

Note:

All models of Deformable DETR are trained with total batch size of 32.
Training and inference speed are measured on NVIDIA Tesla V100 GPU.
"Deformable DETR (single scale)" means only using res5 feature map (of stride 32) as input feature maps for Deformable Transformer Encoder.
"DC5" means removing the stride in C5 stage of ResNet and add a dilation of 2 instead.
"DETR-DC5+" indicates DETR-DC5 with some modifications, including using Focal Loss for bounding box classification and increasing number of object queries to 300.
"Batch Infer Speed" refer to inference with batch size = 4 to maximize GPU utilization.
The original implementation is based on our internal codebase. There are slight differences in the final accuracy and running time due to the plenty details in platform switch.

笔记：

所有可变形DETR的模型都经过训练，总批次大小为32。
训练和推理速度是在NVIDIA Tesla V100 GPU上测量的。
“可变形DETR（单比例）”表示仅将（步幅32的）res5特征图用作可变形变压器编码器的输入特征图。
“ DC5”表示消除ResNet的C5阶段的步幅，而改为增加2。
“ DETR-DC5 +”表示对DETR-DC5进行了一些修改，包括使用Focal Loss进行边界框分类以及将目标查询数增加到300。
“批处理推断速度”指的是批处理大小= 4以最大程度地利用GPU的推理。
原始实现基于内部代码库。由于平台切换器中的大量细节，最终精度和运行时间略有不同。

Installation

Requirements

Linux, CUDA>=9.2, GCC>=5.4
Python>=3.7

We recommend you to use Anaconda to create a conda environment: 建议使用Anaconda创建一个conda环境：

conda create -n deformable_detr python=3.7 pip

Then, activate the environment:

conda activate deformable_detr

PyTorch>=1.5.1, torchvision>=0.6.1

For example, if your CUDA version is 9.2, you could install pytorch and torchvision as following: 如果CUDA版本是9.2，则可以按以下方式安装pytorch和torchvision：

conda install pytorch=1.5.1 torchvision=0.6.1 cudatoolkit=9.2 -c pytorch

Other requirements

pip install -r requirements.txt

Compiling CUDA operators

cd ./models/ops

sh ./make.sh

# unit test (should see all checking is True)

python test.py

Usage

Dataset preparation

Please download COCO 2017 dataset and organize them as following: 请下载COCO 2017数据集并按以下方式组织它们：

code_root/

└── data/

└── coco/

├── train2017/

├── val2017/

└── annotations/

├── instances_train2017.json

└── instances_val2017.json

Training

Training on single node

For example, the command for training Deformable DETR on 8 GPUs is as following: 例如，用于在8个GPU上训练可变形DETR的命令如下：

GPUS_PER_NODE=8 ./tools/run_dist_launch.sh 8 ./configs/r50_deformable_detr.sh

Training on multiple nodes

For example, the command for training Deformable DETR on 2 nodes of each with 8 GPUs is as following: 例如，用于在每个具有8个GPU的2个节点上训练Deformable DETR的命令如下：

On node 1:

MASTER_ADDR=<IP address of node 1> NODE_RANK=0 GPUS_PER_NODE=8 ./tools/run_dist_launch.sh 16 ./configs/r50_deformable_detr.sh

On node 2:

MASTER_ADDR=<IP address of node 1> NODE_RANK=1 GPUS_PER_NODE=8 ./tools/run_dist_launch.sh 16 ./configs/r50_deformable_detr.sh

Training on slurm cluster

If you are using slurm cluster, you can simply run the following command to train on 1 node with 8 GPUs: 如果使用的是Slurm集群，只需运行以下命令即可在具有8个GPU的1个节点上进行训练：

GPUS_PER_NODE=8 ./tools/run_dist_slurm.sh <partition> deformable_detr 8 configs/r50_deformable_detr.sh

Or 2 nodes of each with 8 GPUs:

GPUS_PER_NODE=8 ./tools/run_dist_slurm.sh <partition> deformable_detr 16 configs/r50_deformable_detr.sh

Some tips to speed-up training

If your file system is slow to read images, you may consider enabling '--cache_mode' option to load whole dataset into memory at the beginning of training.

You may increase the batch size to maximize the GPU utilization, according to GPU memory of yours, e.g., set '--batch_size 3' or '--batch_size 4'.
如果文件系统读取图像的速度较慢，则可以考虑在训练开始时启用'--cache_mode'选项以将整个数据集加载到内存中。
可以根据自己的GPU内存来增加批处理大小以最大程度地利用GPU，例如，设置'--batch_size 3'或'--batch_size 4'。

Evaluation

You can get the config file and pretrained model of Deformable DETR (the link is in "Main Results" session), then run following command to evaluate it on COCO 2017 validation set:

可以获取可变形DETR的配置文件和预训练模型（链接在“主要结果”会话中），然后运行以下命令在COCO 2017验证集中对其进行评估：

<path to config file> --resume <path to pre-trained model> --eval

You can also run distributed evaluation by using ./tools/run_dist_launch.sh or ./tools/run_dist_slurm.sh.