Hey, i have created a vehicle speed detection and no. plate capturing system by modifying the code base and fine tuning of YOLO V9 Model.
changed coco.yaml for training
changed yolov9-c.yaml for training
changed yolov9-e.yaml for training
changed utils/general.py line 903 ,, addad [1]
changed detect.py for reading text from liscence during detection
can use docker if want follow the installation guide below if using docker
make virtual env / conda env with python 3.10.11
cloone the repo
go into project folder
install nvidia gpu driver if gpu available (chat gpt se puch lena kaise kar)
pip install -r requirements.txt (do this after removing torch and torchvision from requirements.txt if GPU AVAILABLE)
if gpu not available simple install requirements.txt
install torch Version: 2.1.0+cu118 and torchvision Version: 0.16.0+cu118 if GPU available
pip install "paddleocr>=2.0.1"
pip install prox
pip install common
data
tight
dual
flask
pip install paddle
pip install paddlepaddle
ultralytics
motpy
collections
math
subprocess
csv
change the command in app.py for sourcing and environment activation according to your pc directories
if not using gpu, in the command in app.py , write cpu in place of 0 in --device 0
Install some additional files if they are the dependencies of other installed files (error aa jayega to dikh jayega kya install krna hai)
go into env and python app.py