/seq_nms_yolo_v2

Fork of the seq_nms_yolo repo with better description

Primary LanguageC

Seq_nms_YOLO

Membres: Yunyun SUN, Yutong YAN, Sixiang XU, Heng ZHANG

Introduction

This project combines YOLOv2(reference) and seq-nms(reference) to realise real time video detection.

Steps

  1. Clone the github repository in the folder you would like to have the project:
git clone https://github.com/Ivanioel/seq_nms_yolo_v2
  1. Go inside the project:
cd seq_nms_yolo_v2

  1. Make your proyect using the command make.

    • Take into consideration that there are some flags you should change depending on your environment.
    • GPU=1 # 0 if your pc doesn't support CUDA - If you are going to execute in GPU be sure that the variables COMMON and LDFLAGS are pointing to your cuda installation folder. In this case they are pointing to /usr/local/cuda-10.1/
    • CUDNN=0 # 1 if your pc does support CUDNN
    • OPENCV=0 # 1 if your pc does support OPENCV

  2. Download yolo.weights and tiny-yolo.weights by running:

wget https://pjreddie.com/media/files/yolo.weights
wget https://pjreddie.com/media/files/yolov2-tiny.weights
  1. Create your conda environment and install all the packages required (you have to activate conda first with conda init. For more info https://docs.conda.io/projects/conda/en/latest/user-guide/install/linux.html):
conda create -y -p ./env python=2.7
conda activate ./env

pip install --upgrade tensorflow tf_object_detection
conda install -y opencv matplotlib pillow scipy
  1. Change the path of PKG_CONFIG_PATH (an environment variable that specifies additional paths in which pkg-config will search for its .pc files):
PKG_CONFIG_PATH=$PKG_CONFIG_PATH:$(pwd)/env/lib/pkgconfig
export PKG_CONFIG_PATH
  1. Copy the libdarknet object and link file in your conda environment:
cp libdarknet.so ./env/lib
cp libdarknet.a ./env/lib

  1. Move to the video folder by cd video.

  2. Copy a video file to the video folder, for example, input.mp4

  3. In the video folder run:

python video2img.py -i input.mp4
python get_pkllist.py

  1. Return to project root floder (cd ..) and run python yolo_seqnms.py to generate output images in the video/output folder.

  2. If you want to reconstruct a video from these output images, you can go to the video folder and run python img2video.py -i output

And you will see detection results in video/output

  1. (Extra) If you want to try with different videos and with seq-nms and only nms, execute execute_all.sh in the video folder (giving it execution permission with chmod +x execute_all.sh) and all the sample videos in the video folder will be used.

Reference

This project copies lots of code from darknet , Seq-NMS and models.