Count the number of people crossing a line from a video using an AI model for people detection and tracking.

• People count using line crossing
• Export count to csv

• Clip classification
• Age gender classification
• Person attributes classification
• Google Analytics connection

• Windows, macOS, Linux
• Python 3.7 and later
• ailia SDK 1.2.13 and later
• lap (pip3 install lap)
• ftfy (pip3 install ftfy) if you use clip

1. Put this command to open GUI.

python3 ailia-apps-people-counter.py

2. Push "Input video" button to select input video
3. Push "Set crossing line" button to set crossing line

Click on the screen to draw two lines.

4. Push "Run" button to execute the app

Count (out) is incremented when passing through the line of in to go out. Count (in) is incremented when passing through the line of out to go in.

a. Push "Output video" button to select the path for video b. Push "Output csv" button to select the path for csv c. Push "Output image" button to select the folder for detected image

The examples of csv file.

sec , time, count(in) , count(out) , total_count(in) , total_count(out)
0 , 2023-02-18 12:32:52.289135 , 0 , 0 , 0 , 0
1 , 2023-02-18 12:32:53.794513 , 1 , 1 , 1 , 1
2 , 2023-02-18 12:32:55.195774 , 1 , 1 , 2 , 2

Vehicles can be counted by specifying vehicle as the category. The model must specify yolo. Among yolo categories, count car, truck, and bus as vehicles.

You can send people count information to Google Analytics GA4 using Measurement Protocol.

Connect to Google Analytics

You can set IP camera address from camera settings of menu bar. Please try to set these kind of ip camera address.

rtsp://wowzaec2demo.streamlock.net/vod/mp4:BigBuckBunny_115k.mp4

classDiagram
`ailia APPS People Counter` <|-- `ByteTrack` : Person Tracking (ID assignment)
`ByteTrack` <|-- `ailia.detector` : Person Detection (Bounding box)
`ailia APPS People Counter` : Input Video, Output Video, Output csv
`ByteTrack` : Kalman Filter
`ailia.detector` <|-- `ailia.core` : Inference Engine
`ailia.core` <|-- `onnx` : Model
`ailia.detector` : ailiaDetectorCompute
`ailia.core` : ailiaCreate, ailiaPredict, ailiaDestroy
`ailia.core` <|-- Backend : Acceleration
`onnx` : mot17_s, mot17_tiny
`Backend` : CPU, GPU

Multiple categories can be specified by separating the category names with #.

man,woman#younger,older

Use brew's python3.10.

brew install [email protected]
brew install [email protected]
pip3.10 install -r requrements.txt
python3.10 ailia-people-counter.py

Use python3.10 instead of python3.11.

https://pixabay.com/videos/segway-scooter-people-move-28146/