yuchiwang/py-faster-rcnn-cpu

Faster R-CNN under CPU only mode, using docker under ubuntu 14.04

Faster R-CNN under CPU only mode using docker

About Faster R-CNN, see _README.md

1. System Info

• Ubuntu 14.04 LTS x64 (I'm using VMWare Fusion)
• Note: Please keep memory >= 4GB, or you may encounter following errors:
  • g++: internal compiler error: Killed (program cc1plus) when compiling
  • *ptr host allocation of size 1382400000 failed when running demo.py

2. Install Docker

• see the official document
• set up usergroup

sudo groupadd docker
sudo gpasswd -a ${USER} docker
sudo service docker restart

3. Pull docker image & Install requirements

1. pull docker image

docker pull tleyden5iwx/caffe-cpu-master

2. run image

docker run -i -t tleyden5iwx/caffe-cpu-master /bin/bash

3. install requirements (cython already satisfied)

sudo apt-get install python-opencv
sudo pip install easydict

4. set up git

git config --global user.name "John Doe"
git config --global user.email [email protected]

4. Clone & Run demo

1. clone

git clone --recursive https://github.com/shawnau/py-faster-rcnn-cpu.git

2. compile(assuming the root of the repo is $FRCN_ROOT)

• rename $FRCN_ROOT/caffe-fast-rcnn/Makefile.config.example into Makefile.config

cd $FRCN_ROOT/lib
make

cd $FRCN_ROOT/caffe-fast-rcnn
git checkout faster-rcnn-upstream-33f2445
make -j8 && make pycaffe

3. download pretrained caffemodel

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

4. run demo

cd $FRCN_ROOT
./tools/demo.py --cpu --net zf

• there is a problem with VGG16 net, may be it's memory's issue, see refer

5. results are saved into tag_<filename>.jpgunder $FRCN_ROOT/data/demo/

5. Main modify

1. Problems with the cpu mode

• lib/fast_rcnn/config.py
• lib/fast_rcnn/nms_wrapper.py
• lib/setup.py
• tools/test_net.py, train_net.py
• caffe-fast-rcnn/Makefile.config

2. change output, saving results into files instead of showing

• tools/demo.py

---

Refer

How to setup with CPU ONLY mode
I got the error when running demo.py
Caffe failed with py-faster-rcnn demo.py on TX1