zlingkang / pedestrian_detection_and_tracking

Pedestrian detection using Mobilenet SSD and tracking

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

pedestrian_detection_and_tracking

Pedestrian detection using Mobilenet SSD and tracking

IMAGE ALT TEXT HERE

Pedestrian detection

Trained using Caffe Mobilenet SSD, details(https://github.com/zlingkang/mobilenet_ssd_pedestrian_detection)
Then model transfered to ncnn (a deeplearning framework optimized for mobile platforms)

Tracking

An optical-flow and Kalman Filter based multiple object tracker, more details: https://github.com/zlingkang/multi_object_tracker

Usage

  • Compile ncnn
    go to 3rd_party/ncnn
    (When using with ARM platforms such as Raspberry Pi, modify ncnn/CMakeLists.txt: add add_definitions(-mfpu=neon), modify ncnn/src/CMakeLists.txt: modify line 40to46 to if(TRUE) )
    mkdir build
    cd build
    cmake ..
    make
  • Compile all the rest
    in root directory
    mkdir build
    cd build
    cmake ..
    make
  • Run
    go to build/ ./main

About

Pedestrian detection using Mobilenet SSD and tracking


Languages

Language:C++ 93.6%Language:CMake 6.4%