/pytorch-detect-rfcn

Pytorch implementation of RFCN used as baseline for Imagenet VID+DET in https://arxiv.org/abs/1710.03958.

Primary LanguagePythonMIT LicenseMIT

A pytorch implementation of the baseline
RFCN approach used in the paper https://arxiv.org/abs/1710.03958.

Introduction

This project is a pytorch implementation of the baseline RFCN in the Detect to Track paper. This repository is influenced by the following implementations:

Our implementation stems heavily from the work jwyang/faster-rcnn.pytorch. As in that implementation, this repository has the following qualities:

  • It is pure Pytorch code. We convert all the numpy implementations to pytorch!

  • It supports multi-image batch training. We revise all the layers, including dataloader, rpn, roi-pooling, etc., to support multiple images in each minibatch.

  • It supports multiple GPUs training. We use a multiple GPU wrapper (nn.DataParallel here) to make it flexible to use one or more GPUs, as a merit of the above two features.

  • It is memory efficient. We limit the aspect ratio of the images in each roidb and group images with similar aspect ratios into a minibatch. As such, we can train resnet101 with batchsize = 2 (4 images) on a 2 Titan X (12 GB).

  • Supports 4 pooling methods. roi pooling, roi alignment, roi cropping, and position-sensitive roi pooling. More importantly, we modify all of them to support multi-image batch training.

prerequisites

  • Python 2.7
  • Pytorch 0.3.0 (0.4.0 may work, but hasn't been tested)
  • CUDA 8.0 or higher

Pretrained Model

The RFCN network weights are initialized using the ImageNet resnet-101 weights. The pretrained resnet-101 model can be accessed from here under the name res101.pth

Training

Below are instructions for training an RFCN network on Imagenet VID+DET.

cd pytorch-detect-rfcn
mkdir data

Download the ILSVRC VID and DET (train/val/test lists can be found here. The ILSVRC2015 images can be downloaded from here ).

Untar the file:

tar xf ILSVRC2015.tar.gz

We'll refer to this directory as $DATAPATH. Make sure the directory structure looks something like:

|--ILSVRC2015
|----Annotations
|------DET
|--------train
|--------val
|------VID
|--------train
|--------val
|----Data
|------DET
|--------train
|--------val
|------VID
|--------train
|--------val
|----ImageSets
|------DET
|------VID

Create a soft link under pytorch-detect-rfcn/data:

ln -s $DATAPATH/ILSVRC2015 ./ILSVRC

Create a directory called pytorch-detect-rfcn/data/pretrained_model, and place the pretrained models into this directory.

Before training, set the correct directory to save and load the trained models. The default is ./output/models. Change the arguments "save_dir" and "load_dir" in trainval_net.py and test_net.py to adapt to your environment.

To train an RFCN D&T model with resnet-101 on Imagenet VID, simply run:

CUDA_VISIBLE_DEVICES=$GPU_ID python trainval_net.py \
    --cuda \
    --dataset imagenet_vid \
    --cag \
    --lr $LEARNING_RATE \
    --bs $BATCH_SIZE \

where 'bs' is the batch size with default 1. Above, BATCH_SIZE and WORKER_NUMBER can be set adaptively according to your GPU memory size. On 2 Titan Xps with 12G memory, it can be up to 2 (4 images, 2 per GPU).

Results

Imagenet VID+DET (Train/Test: imagenet_vid_train+imagenet_det_train/imagenet_vid_val, scale=600, PS ROI Pooling).

model   #GPUs batch size lr       lr_decay max_epoch     time/epoch mem/GPU mAP
Res-101     2 2 1e-3 5   11   -- 8021MiB   70.3

Build

As pointed out by ruotianluo/pytorch-faster-rcnn, choose the right -arch to compile the cuda code:

GPU model Architecture
TitanX (Maxwell/Pascal) sm_52
GTX 960M sm_50
GTX 1080 (Ti) sm_61
Grid K520 (AWS g2.2xlarge) sm_30
Tesla K80 (AWS p2.xlarge) sm_37

More details about setting the architecture can be found here or here

Install all the python dependencies using pip:

pip install -r requirements.txt

Compile the cuda dependencies using following simple commands:

cd lib
sh make.sh

It will compile all the modules you need, including NMS, PSROI_POOLING, ROI_Pooing, ROI_Align and ROI_Crop. The default version is compiled with Python 2.7, please compile by yourself if you are using a different python version.

As pointed out in this issue, if you encounter some error during the compilation, you might miss to export the CUDA paths to your environment.

Authorship

Contributions to this project have been made by Thomas Balestri and Jugal Sheth.