The paper presents a globally optimal greedy algorithm for multi-object tracking in video sequences. The authors formulate the problem using a cost function that estimates the number of tracks and their birth and death states. They show that the global solution can be obtained using a greedy algorithm that sequentially instantiates tracks through shortest path computations on a flow network. This approach allows for the integration of pre-processing steps, such as non-maximum suppression (NMS), within the tracking algorithm. The paper introduces an approximate greedy algorithm with linear time complexity in the number of objects and video length, which is significantly faster than previous methods. The algorithms are evaluated on benchmark datasets, demonstrating state-of-the-art performance in terms of detection accuracy and track identity. The greedy algorithms are scalable and can handle dense input data, making them suitable for real-world applications.The paper presents a globally optimal greedy algorithm for multi-object tracking in video sequences. The authors formulate the problem using a cost function that estimates the number of tracks and their birth and death states. They show that the global solution can be obtained using a greedy algorithm that sequentially instantiates tracks through shortest path computations on a flow network. This approach allows for the integration of pre-processing steps, such as non-maximum suppression (NMS), within the tracking algorithm. The paper introduces an approximate greedy algorithm with linear time complexity in the number of objects and video length, which is significantly faster than previous methods. The algorithms are evaluated on benchmark datasets, demonstrating state-of-the-art performance in terms of detection accuracy and track identity. The greedy algorithms are scalable and can handle dense input data, making them suitable for real-world applications.