No module named 'MultiScaleDeformableAttention'

Deformable DETR

By Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.

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

Introduction

TL; DR. 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.

Abstract. 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.

License

This project is released under the Apache 2.0 license.

Changelog

See changelog.md for detailed logs of major changes.

Citing Deformable DETR

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

@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

_{_Method}	_{_Epochs}	_{_AP}	_{_APS}	_{_APM}	_{_APL}	_{_params (M)}	_{_FLOPs (G)}	_{_{Total Train Time (GPU hours)}}	_{_{Train Speed (GPU hours /epoch)}}	_{_{Infer Speed (FPS)}}	_{_{Batch Infer Speed (FPS)}}	_{_URL}
_{_{Faster R-CNN + FPN}}	₁₀₉	_42.0	_26.6	_45.4	_53.4	₄₂	₁₈₀	₃₈₀	_3.5	_25.6	_28.0	_-
_{_DETR}	₅₀₀	_42.0	_20.5	_45.8	_61.1	₄₁	₈₆	₂₀₀₀	_4.0	_27.0	_38.3	_-
_{_DETR-DC5}	₅₀₀	_43.3	_22.5	_47.3	_61.1	₄₁	₁₈₇	₇₀₀₀	_14.0	_11.4	_12.4	_-
_{_DETR-DC5}	₅₀	_35.3	_15.2	_37.5	_53.6	₄₁	₁₈₇	₇₀₀	_14.0	_11.4	_12.4	_-
_{_DETR-DC5+}	₅₀	_36.2	_16.3	_39.2	_53.9	₄₁	₁₈₇	₇₀₀	_14.0	_11.4	_12.4	_-
_{_{Deformable DETR (single scale)}}	₅₀	_39.4	_20.6	_43.0	_55.5	₃₄	₇₈	₁₆₀	_3.2	_27.0	_42.4	_{config log model}
_{_{Deformable DETR (single scale, DC5)}}	₅₀	_41.5	_24.1	_45.3	_56.0	₃₄	₁₂₈	₂₁₅	_4.3	_22.1	_29.4	_{config log model}
_{_{Deformable DETR}}	₅₀	_44.5	_27.1	_47.6	_59.6	₄₀	₁₇₃	₃₂₅	_6.5	_15.0	_19.4	_{config log model}
_{_{+ iterative bounding box refinement}}	₅₀	_46.2	_28.3	_49.2	_61.5	₄₁	₁₇₃	₃₂₅	_6.5	_15.0	_19.4	_{config log model}
_{_{++ two-stage Deformable DETR}}	₅₀	_46.9	_29.6	_50.1	_61.6	₄₁	₁₇₃	₃₄₀	_6.8	_14.5	_18.8	_{config log model}

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.

Installation

Requirements

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

We recommend you to use Anaconda to create a conda environment:
```
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 (following instructions here)

For example, if your CUDA version is 9.2, you could install pytorch and torchvision as following:
```
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:

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:

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:

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:

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'.

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:

<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.

	# for samples, targets in metric_logger.log_every(data_loader, print_freq, header):
	for _ in metric_logger.log_every(range(len(data_loader)), print_freq, header):
	outputs = model(samples)
	loss_dict = criterion(outputs, targets)
	weight_dict = criterion.weight_dict
	losses = sum(loss_dict[k] * weight_dict[k] for k in loss_dict.keys() if k in weight_dict)

	# reduce losses over all GPUs for logging purposes
	loss_dict_reduced = utils.reduce_dict(loss_dict)
	loss_dict_reduced_unscaled = {f'{k}_unscaled': v
	for k, v in loss_dict_reduced.items()}
	loss_dict_reduced_scaled = {k: v * weight_dict[k]
	for k, v in loss_dict_reduced.items() if k in weight_dict}
	losses_reduced_scaled = sum(loss_dict_reduced_scaled.values())

	loss_value = losses_reduced_scaled.item()

	if not math.isfinite(loss_value):
	print("Loss is {}, stopping training".format(loss_value))
	print(loss_dict_reduced)
	sys.exit(1)

	optimizer.zero_grad()
	losses.backward()
	if max_norm > 0:
	grad_total_norm = torch.nn.utils.clip_grad_norm_(model.parameters(), max_norm)
	else:
	grad_total_norm = utils.get_total_grad_norm(model.parameters(), max_norm)
	optimizer.step()

fundamentalvision / deformable-detr Goto Github PK

deformable-detr's Introduction

Deformable DETR

Introduction

License

Changelog

Citing Deformable DETR

Main Results

Installation

Requirements

Compiling CUDA operators

Usage

Dataset preparation

Training

Training on single node

Training on multiple nodes

Training on slurm cluster

Some tips to speed-up training

Evaluation

deformable-detr's People

Contributors

Stargazers

Watchers

Forkers

deformable-detr's Issues

Recommend Projects

Recommend Topics

Recommend Org