Steps to use YOLOv5

Python Packages Dependencies

  • Install following python paackages

    • matplotlib>=3.2.2

    • numpy>=1.18.5

    • opencv-python>=4.1.2

    • Pillow>=7.1.2

    • PyYAML>=5.3.1

    • requests>=2.23.0

    • scipy>=1.4.1

    • torch>=1.7.0

    • torchdivision>=0.8.1

    • tqdm>=4.41.0

    • tensorboard>=2.4.1

    • pandas>=1.1.4

    • seaborn>=0.11.0

Environment Setup for YOLOv5

  1. Clone YOLOv5 repository
  2. Go to the yolov5 directory

    • cd yolov5

  3. Install python libraries using

    • pip install -r requirements.txt

Run YOLOv5 Model

  1. Download pretrained yolov5 weights from the link and save it in the sae directory ‘yolov5’
  2. For real time detection execute the commands

    • python3 detect.py –weights yolov5s.pt –source 0

  3. To run the model on the downloded video or images download the video or the image and put it in the yolov5 folder and then use the command as

    • For Image : python3 detect.py –weights yolov5s.pt –source imgname.jpg

    • For Video : python3 detect.py –weights yolov5s.pt –source videofilename.mp4

Issue faced

The installation command “pip install -r requirements.txt” works well with the jetpack 4.6 but in jetpack 4.4 the command fails while installing Opencv as can be seen in the screenshot below:

../_images/yolov5_opencv_installation_issue.png

Solution

The issue can be resolved upgarding the version of pip, as the pip version installed with jetpack 4.4 is 9.0.1 and it needs higher version of pip. pip version can be upgarded using command as:

  • pip3 install –upgrade pip

Note

  • yolov5s.pt is PyTorch format of Yolov5. Other supported formats like TorchScript, TensorRT can also be used with the same command just replacing the correspoding weightfiles format

  • python3 detect.py –weights weightfiles.xx –source 0