This paper presents a simple and efficient online tracking framework for multiple object tracking (MOT), focusing on frame-to-frame prediction and association. The key idea is that detection quality significantly impacts tracking performance, and by leveraging recent advancements in detection, state-of-the-art tracking accuracy can be achieved with classical tracking methods. The proposed method uses a Kalman filter for motion prediction and the Hungarian algorithm for data association, achieving a tracking rate of 260 Hz, which is over 20 times faster than other state-of-the-art trackers. The method ignores complex features beyond detection and focuses on bounding box position and size for motion estimation and data association. It also ignores short-term and long-term occlusion as they are rare and introduce unnecessary complexity. The method is evaluated on the MOT benchmark, showing that it achieves the highest MOTA score among online trackers and is comparable to the more complex NOMT method. The framework is simple and efficient, making it suitable as a baseline for research and applications such as collision avoidance. The method uses a CNN-based detector (FrRCNN) for object detection, and the results show that the best detector (FrRCNN(VGG16)) leads to the best tracking accuracy. The framework is open-sourced to facilitate research and experimentation. The paper concludes that detection quality is crucial for tracking performance and that future work should explore tightly coupled detection and tracking frameworks.This paper presents a simple and efficient online tracking framework for multiple object tracking (MOT), focusing on frame-to-frame prediction and association. The key idea is that detection quality significantly impacts tracking performance, and by leveraging recent advancements in detection, state-of-the-art tracking accuracy can be achieved with classical tracking methods. The proposed method uses a Kalman filter for motion prediction and the Hungarian algorithm for data association, achieving a tracking rate of 260 Hz, which is over 20 times faster than other state-of-the-art trackers. The method ignores complex features beyond detection and focuses on bounding box position and size for motion estimation and data association. It also ignores short-term and long-term occlusion as they are rare and introduce unnecessary complexity. The method is evaluated on the MOT benchmark, showing that it achieves the highest MOTA score among online trackers and is comparable to the more complex NOMT method. The framework is simple and efficient, making it suitable as a baseline for research and applications such as collision avoidance. The method uses a CNN-based detector (FrRCNN) for object detection, and the results show that the best detector (FrRCNN(VGG16)) leads to the best tracking accuracy. The framework is open-sourced to facilitate research and experimentation. The paper concludes that detection quality is crucial for tracking performance and that future work should explore tightly coupled detection and tracking frameworks.