SFND 2D Feature Tracking

The idea of the camera course is to build a collision detection system - that's the overall goal for the Final Project. As a preparation for this, we build the feature tracking part and test various detector / descriptor combinations to see which ones perform best. This project consists of four parts:

• First, loading images, setting up data structures and putting everything into a ring buffer to optimize memory load.
• Then, integrating several keypoint detectors such as HARRIS, FAST, BRISK and SIFT and comparing them with regard to number of keypoints and speed.
• The next part, descriptor extraction and matching using brute force and also the FLANN approach.
• The last part, once the code framework is complete, various algorithms in different combinations are being tested and they are compared with regard to some performance measures.

Dependencies for Running Locally

• cmake >= 2.8
  • All OSes: click here for installation instructions
• make >= 4.1 (Linux, Mac), 3.81 (Windows)
  • Linux: make is installed by default on most Linux distros
  • Mac: install Xcode command line tools to get make
  • Windows: Click here for installation instructions
• OpenCV >= 4.1
  • This must be compiled from source using the -D OPENCV_ENABLE_NONFREE=ON cmake flag for testing the SIFT and SURF detectors.
  • The OpenCV 4.1.0 source code can be found here
• gcc/g++ >= 5.4
  • Linux: gcc / g++ is installed by default on most Linux distros
  • Mac: same deal as make - install Xcode command line tools
  • Windows: recommend using MinGW
• for argument parsing I am using argparse in this github repo which is already included as a submodule here.

Basic Build Instructions

1. Clone this repo and submodule argparse as follow.

   git clone https://github.com/yosoufe/SFND_2D_Feature_Tracking.git
   cd SFND_2D_Feature_Tracking
   git submodule update --init --recursive
   

2. Make a build directory in the top level directory: mkdir build && cd build

3. Compile without GPU Support: cmake .. && make

   • With GPU support: cmake -DWITH_CUDA=ON .. && make:

     In case with of GPU support the extra following items would be available:

     • detectors: ORB_CUDA and FAST_CUDA
     • matcher: MAT_BF_CUDA
     • descriptor: ORB_CUDA

4. Run it:

   • The executable accepts multiple optional arguments to define multiple variables.

   • use ./2D_feature_tracking -h for help. It would create the following output:

     $ ./2D_feature_tracking -h
     Usage: ./2D_feature_tracking [args]
     For example: ./2D_feature_tracking --detector_type=BRISK --matcher_type=MAT_FLANN --descriptor_type=DES_BINARY --selector_type=SEL_KNN -f -q
     
     Explores different 2d keypoint detector, descriptor and matching
     
         -h, --help                show this help message and exit
     
     Optional Arguments: 
         --detector_type=<str>     detector type, options: SHITOMASI, HARRIS, FAST, BRISK, ORB, AKAZE, and SIFT
                                   if compiled (WITH_CUDA on): ORB_CUDA, FAST_CUDA
                                   default: SHITOMASI
         --matcher_type=<str>      matcher type, options: MAT_BF, MAT_FLANN,
                                   if compiled (WITH_CUDA on): MAT_BF_CUDA
                                   default: MAT_BF
         --descriptor_type=<str>   descriptor type, options: BRISK BRIEF, ORB, FREAK, AKAZE, SIFT
                                   if compiled (WITH_CUDA on): ORB_CUDA
                                   default: BRISK
         --selector_type=<str>     selector type, options: SEL_NN, SEL_KNN
                                   default: SEL_NN
         -f, --focus_on_vehicle    To focus on only keypoints that are on the preceding vehicle.
         -l, --limit_keypoints     To limit the number of keypoints to maximum 50 keypoints.
         -q, --quiet               If this flag is chosen no image would be shown. Good for performance measurement