Yolov5_DeepSort_Pytorch

Yolov5_DeepSort_Pytorch

Product Information

favorites Add to favorites

Public chat

Support Plans

There are currently no OSS plans available

If you are a provider or contributor to the repository, you can start adding your OSS plan.

Add an OSS plan
New to PieceX for OSS? Learn more here

Contact us if you are looking for a plan for this open source.

We will help you get in touch with professional providers.

Product Details

Real-time multi-object tracker using YOLO v5 and deep sort

Easy training program of Yolov5 Deep Sort with PyTorch

Prerequisites

  • Already installed Docker

How to use

  • create environment (common step)

    • execute following script in command line
      
      # clone this repository
      git clone --recurse-submodules https://github.com/tmfi-analytics/Yolov5_DeepSort_Pytorch.git

    build Docker container using Dockerfile

    cd Yolov5_DeepSort_Pytorch docker build -t [YOUR CONTAINER NAME] .

3 ways to use

  • Just execute human tracking demo

    • edit config.sh as following

      # 訓練済みのモデルで物体追跡をする場合はここを1に設定
      JUST_PREDICTION=1
      
      # YOLOV5の訓練に自前のデータセットを使用する場合はここを1に設定
      USE_CUSTOM_DATASET=0
    • execute docker run command
      cd Yolov5_DeepSort_Pytorch
      docker run -it --env-file=config.sh -v `pwd`/data:/app/data --shm-size=2048m [YOUR CONTAINER NAME]
    • watch the output video
      ls data/result
  • train the human detection model and track using prepared dataset

  • train the human detection model and track using custom dataset

    • put train and val dataset as following
        スクリーンショット 2021-12-14 23 36 52
      put image files in images directory and label files in labels directory. the pair of image file and label file must be same file name
      please refer the following link about label format
      https://github.com/ultralytics/yolov5/wiki/Train-Custom-Data
    • put mp4 file for test as data/test/test.mp4
    • edit config.sh as following

      # 訓練済みのモデルで物体追跡をする場合はここを1に設定
      JUST_PREDICTION=0
      
      # YOLOV5の訓練に自前のデータセットを使用する場合はここを1に設定
      USE_CUSTOM_DATASET=1
    • execute docker run command
      cd Yolov5_DeepSort_Pytorch
      docker run -it --env-file=config.sh -v `pwd`/data:/app/data --shm-size=2048m [YOUR CONTAINER NAME]
    • watch the output video
      ls data/result
user-symbol

Stay in touch

Get Practical Tips For Business and Developers.

Learn about PieceX Community Needs for Source Code Projects.

Be the First to Know PieceX Newest Free Community Code Projects.