People Counting with YOLO & DeepSORT

This project is a Computer Vision task for people counting using YOLO for object detection and DeepSORT for object tracking. The implementation is provided in a Jupyter Notebook.

📌 Features

• Detects people in video streams or images using YOLO (Ultralytics).

• Tracks detected people across frames using DeepSORT.

• Counts the number of people in the frame over time.

• Runs fully on pre-trained YOLO weights (no additional dataset required).

📦 Dependencies

Make sure you have the following Python libraries installed:

pip install ultralytics deep_sort_realtime pillow numpy opencv-python

Additionally, the following built-in Python libraries are used:

shutil
warnings
cProfile
pstats
io
argparse

🚀 Usage

1- Open the Notebook

• Launch Jupyter Notebook or JupyterLab:

jupyter notebook

• Open Acacus_Internship_Task.ipynb.

2- Run the Cells

• The notebook will:

  1- Load YOLO and DeepSORT.

  2- Process input video or webcam.

  3- Display the tracking and counting results in real-time or save to a file.

3- Change Input Source

• Update the notebook’s input path to your desired video file or webcam index (e.g., 0 for default webcam).

📂 Project Structure

Acacus_Internship_Task.ipynb   # Main notebook

📊 Output

• Annotated video with bounding boxes and tracking IDs.

• Console or cell output with people count.

⚡ Notes

• Ensure that your environment has GPU acceleration for faster inference.

• You can adjust YOLO and DeepSORT parameters (like confidence thresholds) directly in the notebook.

• For larger videos, consider profiling using cProfile to optimize performance.