Home

Awesome

This code base is no longer maintained and exists as a historical artifact to supplement my ICCV 2015 paper. For more recent work that's faster and more accurrate, please see Faster R-CNN (which also includes functionality for training Fast R-CNN).

Fast R-CNN: Fast Region-based Convolutional Networks for object detection

Created by Ross Girshick at Microsoft Research, Redmond.

Modified by Peng Tang to adapt training fast-rcnn by pseudo ground truths.

The pseudo should be putted under the folder data/selective_search_data, and named as voc_2007_trainval_gt.mat (or voc_2012_trainval_gt.mat).

Introduction

Fast R-CNN is a fast framework for object detection with deep ConvNets. Fast R-CNN

Fast R-CNN was initially described in an arXiv tech report and later published at ICCV 2015.

License

Fast R-CNN is released under the MIT License (refer to the LICENSE file for details).

Citing Fast R-CNN

If you find Fast R-CNN useful in your research, please consider citing:

@inproceedings{girshickICCV15fastrcnn,
    Author = {Ross Girshick},
    Title = {Fast R-CNN},
    Booktitle = {International Conference on Computer Vision ({ICCV})},
    Year = {2015}
}

Contents

  1. Requirements: software
  2. Requirements: hardware
  3. Basic installation
  4. Demo
  5. Beyond the demo: training and testing
  6. Usage
  7. Extra downloads

Requirements: software

  1. Requirements for Caffe and pycaffe (see: Caffe installation instructions)

Note: Caffe must be built with support for Python layers!

# In your Makefile.config, make sure to have this line uncommented
WITH_PYTHON_LAYER := 1

You can download my Makefile.config for reference. 2. Python packages you might not have: cython, python-opencv, easydict 3. [optional] MATLAB (required for PASCAL VOC evaluation only)

Requirements: hardware

  1. For training smaller networks (CaffeNet, VGG_CNN_M_1024) a good GPU (e.g., Titan, K20, K40, ...) with at least 3G of memory suffices
  2. For training with VGG16, you'll need a K40 (~11G of memory)

Installation (sufficient for the demo)

  1. Clone the Fast R-CNN repository
# Make sure to clone with --recursive
git clone --recursive https://github.com/ppengtang/fast-rcnn.git

  1. We'll call the directory that you cloned Fast R-CNN into FRCN_ROOT

    Ignore notes 1 and 2 if you followed step 1 above.

    Note 1: If you didn't clone Fast R-CNN with the --recursive flag, then you'll need to manually clone the caffe-fast-rcnn submodule:

    git submodule update --init --recursive

    Note 2: The caffe-fast-rcnn submodule needs to be on the fast-rcnn branch (or equivalent detached state). This will happen automatically if you follow these instructions.

  2. Build the Cython modules

    cd $FRCN_ROOT/lib
make

  3. Build Caffe and pycaffe

    cd $FRCN_ROOT/caffe-fast-rcnn
# Now follow the Caffe installation instructions here:
#   http://caffe.berkeleyvision.org/installation.html

# If you're experienced with Caffe and have all of the requirements installed
# and your Makefile.config in place, then simply do:
make -j8 && make pycaffe

  4. Download pre-computed Fast R-CNN detectors

    cd $FRCN_ROOT
./data/scripts/fetch_fast_rcnn_models.sh

    This will populate the $FRCN_ROOT/data folder with fast_rcnn_models. See data/README.md for details.

Demo

After successfully completing basic installation, you'll be ready to run the demo.

Python

To run the demo

cd $FRCN_ROOT
./tools/demo.py

The demo performs detection using a VGG16 network trained for detection on PASCAL VOC 2007. The object proposals are pre-computed in order to reduce installation requirements.

Note: If the demo crashes Caffe because your GPU doesn't have enough memory, try running the demo with a small network, e.g., ./tools/demo.py --net caffenet or with --net vgg_cnn_m_1024. Or run in CPU mode ./tools/demo.py --cpu. Type ./tools/demo.py -h for usage.

MATLAB

There's also a basic MATLAB demo, though it's missing some minor bells and whistles compared to the Python version.

cd $FRCN_ROOT/matlab
matlab # wait for matlab to start...

# At the matlab prompt, run the script:
>> fast_rcnn_demo

Fast R-CNN training is implemented in Python only, but test-time detection functionality also exists in MATLAB. See matlab/fast_rcnn_demo.m and matlab/fast_rcnn_im_detect.m for details.

Computing object proposals

The demo uses pre-computed selective search proposals computed with this code. If you'd like to compute proposals on your own images, there are many options. Here are some pointers; if you run into trouble using these resources please direct questions to the respective authors.

  1. Selective Search: original matlab code, python wrapper
  2. EdgeBoxes: matlab code
  3. GOP and LPO: python code
  4. MCG: matlab code
  5. RIGOR: matlab code

Apologies if I've left your method off this list. Feel free to contact me and ask for it to be included.

Beyond the demo: installation for training and testing models

  1. Download the training, validation, test data and VOCdevkit

    wget http://host.robots.ox.ac.uk/pascal/VOC/voc2007/VOCtrainval_06-Nov-2007.tar
wget http://host.robots.ox.ac.uk/pascal/VOC/voc2007/VOCtest_06-Nov-2007.tar
wget http://host.robots.ox.ac.uk/pascal/VOC/voc2007/VOCdevkit_08-Jun-2007.tar

  2. Extract all of these tars into one directory named VOCdevkit

    tar xvf VOCtrainval_06-Nov-2007.tar
tar xvf VOCtest_06-Nov-2007.tar
tar xvf VOCdevkit_08-Jun-2007.tar

  3. It should have this basic structure

    $VOCdevkit/                           # development kit
$VOCdevkit/VOCcode/                   # VOC utility code
$VOCdevkit/VOC2007                    # image sets, annotations, etc.
# ... and several other directories ...

  4. Create symlinks for the PASCAL VOC dataset

    cd $FRCN_ROOT/data
ln -s $VOCdevkit VOCdevkit2007

    Using symlinks is a good idea because you will likely want to share the same PASCAL dataset installation between multiple projects.

  5. [Optional] follow similar steps to get PASCAL VOC 2010 and 2012

  6. Follow the next sections to download pre-computed object proposals and pre-trained ImageNet models

Download pre-computed Selective Search object proposals

Pre-computed selective search boxes can also be downloaded for VOC2007 and VOC2012.

cd $FRCN_ROOT
./data/scripts/fetch_selective_search_data.sh

This will populate the $FRCN_ROOT/data folder with selective_selective_data.

Download pre-trained ImageNet models

Pre-trained ImageNet models can be downloaded for the three networks described in the paper: CaffeNet (model S), VGG_CNN_M_1024 (model M), and VGG16 (model L).

cd $FRCN_ROOT
./data/scripts/fetch_imagenet_models.sh

These models are all available in the Caffe Model Zoo, but are provided here for your convenience.

Usage

Train a Fast R-CNN detector. For example, train a VGG16 network on VOC 2007 trainval:

./tools/train_net.py --gpu 0 --solver models/VGG16/solver.prototxt \
	--weights data/imagenet_models/VGG16.v2.caffemodel

If you see this error

EnvironmentError: MATLAB command 'matlab' not found. Please add 'matlab' to your PATH.

then you need to make sure the matlab binary is in your $PATH. MATLAB is currently required for PASCAL VOC evaluation.

Test a Fast R-CNN detector. For example, test the VGG 16 network on VOC 2007 test:

./tools/test_net.py --gpu 1 --def models/VGG16/test.prototxt \
	--net output/default/voc_2007_trainval/vgg16_fast_rcnn_iter_40000.caffemodel

Test output is written underneath $FRCN_ROOT/output.

Compress a Fast R-CNN model using truncated SVD on the fully-connected layers:

./tools/compress_net.py --def models/VGG16/test.prototxt \
	--def-svd models/VGG16/compressed/test.prototxt \
    --net output/default/voc_2007_trainval/vgg16_fast_rcnn_iter_40000.caffemodel
# Test the model you just compressed
./tools/test_net.py --gpu 0 --def models/VGG16/compressed/test.prototxt \
	--net output/default/voc_2007_trainval/vgg16_fast_rcnn_iter_40000_svd_fc6_1024_fc7_256.caffemodel

Experiment scripts

Scripts to reproduce the experiments in the paper (up to stochastic variation) are provided in $FRCN_ROOT/experiments/scripts. Log files for experiments are located in experiments/logs.

Note: Until recently (commit a566e39), the RNG seed for Caffe was not fixed during training. Now it's fixed, unless train_net.py is called with the --rand flag. Results generated before this commit will have some stochastic variation.

Extra downloads