Project information

• Category: AI
• Project date: September, 2023
• Project URL: GitHub Repository

Project description

In this project, we use a machine learning model to detect and track the humans in a part of the 1917 movie, about 90 seconds. There are different approaches to this problem, and many people have implemented other methods to achieve this goal. Usually, a combination of an object detection model and a tracking algorithm is used to detect and track objects. One of the state-of-the-art methods is Yolo, a popular object detection model known for its speed and accuracy.

The latest version of the Yolo family, known as Yoplov8, can detect and track objects by itself. Yolov8 has various models with different numbers of parameters, leading to different speeds and mean average precision (mAP). Here, we use the lightest version (yolov8n) to fine-tune for human tracking tasks. The results show that the fine-tuned model is more accurate than the original in human detection task.