Vehicle Detection and Counting System on Streamlit

Introduction

This project is used to count and detect vehicle on the highway. It can detect 4 types of vehicles: car, motorcycle, bus, truck. I run this project on Python 3.9.7

• YOLOv5 to detect objects on each of the video frames.

• Deep SORT to track those objects over different frames and help counting.

• Streamlit to build a simple web.

Installation

• Install essential libraries and packages:

pip install -r requirements.txt

• Run demo:

streamlit run demo.py --server.maxUploadSize=500

NOTE: If the web keeps showing "Please wait...", try to install streamlit version 1.11.0

pip install streamlit==1.11.0

If the web shows error "no module easydict"

pip install easydict

DEMO

Steps:

1. Click Browse files to input video

2. Setting Custom classes, Confidence and Line position

• Custom classes: choose classes you want to detect

• Confidence: the probability that one object belongs to one class

• Line position: the position of green line, any vehicle have coordinate below the line will be counted

3. Click START

Result