Resources

• How to Use DynaSLAM using Docker? DynaSLAM安装与运行
• DynaSLAM: Tracking, Mapping and Inpainting in Dynamic Scenes
• DynaSLAM 代码分析

Video Demo

https://www.bilibili.com/video/av77344939

Dependences

• python: 2.7
• tensorflow
• keras: 2.0.8
• cuda: 9.0
• OpenCV: 2.4.13

Prepare

Kick off docker

git clone https://github.com/yubaoliu/DynaSLAM.git
cd DynaSLAM/docker
sudo docker-compose build
sudo docker-compose up

Build DynaSLAM

cd DynaSLAM
chmod +x build.sh
./build.sh

RUN

Mount dataset folder

vim docker/.env: Modify DATASET_DIR

Get module

echo "get trained model"
cd DynaSLAM/src/python/
wget -c https://github.com/matterport/Mask_RCNN/releases/download/v1.0/mask_rcnn_coco.h5
wget -c  https://github.com/matterport/Mask_RCNN/releases/download/v2.1/mask_rcnn_balloon.h5 

Kick of docker container

# xhost 控制X server访问权限
xhost +local:root
sudo docker-compose run dynaslam bash

Start DynaSLAM

check run_in_docker.sh

Results

Test

Test tensorflow

import tensorflow as tf
print(tf.__version__)

Test or Check MaskRCNN

DynaSLAM$ python src/python/Check.py

If everything works fun:

::::
Initialated Mask RCNN network...
Mask R-CNN is correctly working

Tips when building

libprotobuf

Problem:

[libprotobuf FATAL google/protobuf/stubs/common.cc:61] This program requires version 3.6.0 of the Protocol Buffer runtime library, but the installed version is 2.6.1.  Please update your library.  If you compiled the program yourself, make sure that your headers are from the same version of Protocol Buffers as your link-time library.  (Version verification failed in "external/protobuf_archive/src/google/protobuf/any.pb.cc".)
terminate called after throwing an instance of 'google::protobuf::FatalException'
  what():  This program requires version 3.6.0 of the Protocol Buffer runtime library, but the installed version is 2.6.1.  Please update your library.  If you compiled the program yourself, make sure that your headers are from the same version of Protocol Buffers as your link-time library.  (Version verification failed in "external/protobuf_archive/src/google/protobuf/any.pb.cc".)
Aborted (core dumped)

Check Version:

protoc --version

ndarrayobject.h: No such file or directory

In file included from /root/DynaSlam/src/Conversion.cc:9:0:
/root/DynaSlam/include/Conversion.h:17:27: fatal error: ndarrayobject.h: No such file or directory
compilation terminated.
CMakeFiles/DynaSLAM.dir/build.make:518: recipe for target 'CMakeFiles/DynaSLAM.dir/src/Conversion.cc.o' failed
make[2]: *** [CMakeFiles/DynaSLAM.dir/src/Conversion.cc.o] Error 1
make[2]: *** Waiting for unfinished jobs....
In file included from /root/DynaSlam/include/MaskNet.h:26:0,
                 from /root/DynaSlam/src/MaskNet.cc:9:
/root/DynaSlam/include/Conversion.h:17:27: fatal error: ndarrayobject.h: No such file or directory
compilation terminated.

solution:

root@710c930fa912:~/DynaSlam/build# find / -name arrayobject.h
/usr/local/lib/python2.7/dist-packages/numpy/core/include/numpy/arrayobject.h

OLD README

Original README.md -> README.md.old

Another method

docker pull yubaoliu/cuda-9-cv2:Mask-RCNN

docker run -it --rm \
    -v /home/yubao/data/share:/root \
    -v /home/yubao/data/Dataset:/Dataset \
    --net host \
    --runtime=nvidia \
    -e DISPLAY=$$DISPLAY \
    -v $$HOME/.Xauthority:/root/.Xauthority \
    --name cuda9cv2 \
    yubaoliu/cuda-9-cv2:Mask-RCNN