This is the repository of Megvii Challenge in paper re-implementation.
This work is based on IDM: An Intermediate Domain Module for Domain Adaptive Person Re-ID.
Market1501 by Tsinghua: Inside the dataset, there are five folders -- bounding_box_test, bounding_box_train, gt_bbox, gt_query and query. The bounding_box_test is your gallery, where there is good detection and could be a match to the query and irrelevant detection. The gt_bbox is the manually assigned bounding box and used to test the quality of the bounding box. gt_query is used to test which is a good match (same person within/across cameras) and which is a bad match (different person within camera / across camera). query has six images for each person and no repetition to test set.
DUKEMTMC by Duke: Inside the dataset, there are three folders -- bounding_box_test, bounding_box_train, query. Like a simplified version of Market1501. query has three images for each person. It's simpler since a few of persons are shot under only one camera and there is no irrelevant detection.