NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:209
Title:muSSP: Efficient Min-cost Flow Algorithm for Multi-object Tracking

• The paper presents a method on efficient min-cost flow algorithm for the multi-object tracking problem. The main idea of the min-cost flow algorithm is to remove unnecessary edges in the graph in order to find a more efficient solution in term of time. Then through a series of operation on the shortest path tree, it poposes a good solutions. Initially reviewers disagree in the rate with a very satisfied reviewer ( rating it 9), a very negative one ( rating it 4) and a positive reviewer . This reviewer was satisfied on the rebuttal rating it 7 at the end. One reviewer remains not convinced since in the paper there are no evaluation in term of accuracy that is a strong point for multiple-target tracking. After the discussion the area chair agrees for the acceptance