/FAST_ARM64

Experimental implementation of FAST Corner Detection used for YGZ-Stereo-Inertial SLAM (Based on ORBSLAM)

Primary LanguageC++GNU Lesser General Public License v2.1LGPL-2.1

FAST corner detector by Edward Rosten. The sourcefiles are from the CVD library: http://www.edwardrosten.com/cvd/

This repository is an experimental implementation of FAST for ARM NEON compilers.

Made possible by https://github.com/DLTcollab/sse2neon

This package is required by https://github.com/gaoxiang12/ygz-stereo-inertial However in order to install this package on a NVIDIA Jetson Device it must be ARM compatible.

git clone

cd FAST_ARM64

mkdir build

cd /build

cmake ..

make -j4

make install -j4

cd ../test

EDIT LINE 10 with path to image in /data/test1.png

g++ -I/usr/local/include/opencv -I/usr/local/include/opencv2 -I/usr/local/include/fast -L/usr/local/lib/ -g -o testyboy test.cpp -lopencv_core -lopencv_imgproc -lopencv_highgui -lopencv_ml -lopencv_video -lopencv_features2d -lopencv_calib3d -lopencv_objdetect -lopencv_stitching -lopencv_imgcodecs -lfast

note if you installed your own opencv from source (presumably people following the GAAS project would have done this)

I ran into this issue: https://github.com/cggos/dip_cvqt/issues/1

After following the steps for creating a .conf file I was able to run </sudo ldconfig -v> and my opencv_core was able to link with the shared libraries directory

chmod +x testyboy

./testyboy

Testing PLAIN version

PLAIN took 1.522497 ms (average over 1000 trials).

PLAIN version extracted 167 features.

Testing Modified ARM version

Modified ARM version took 1.458408 ms (average over 1000 trials).

Modified ARM version extracted 167 features.

BENCHMARK version extracted 167 features.