SYSU-30k Dataset of "Weakly Supervised Person Re-ID: Differentiable Graphical Learning and A New Benchmark" https://arxiv.org/abs/1904.03845
SYSU-30k contains 30k categories of persons, which is about 20 times large rthan CUHK03 (1.3k categories)and Market1501 (1.5k categories), and 30 times larger than ImageNet (1k categories). SYSU-30k contains 29,606,918 images. Moreover, SYSU-30k provides not only a large platform for the weakly supervised ReID problem but also a more challenging test set that is consistent with the realistic setting for standard evaluation. Figure 1 shows some samples from the SYSU-30k dataset.
Figure 1: Examples in the new SYSU-30k dataset. (a) Person images in terms of bag and (b) their bag-level weak annotations.
Figure 2: The statistics of the SYSU-30k. (a) summarizes the number of bags with respect to the number of images per bag. (b) and (c) compare SYSU-30k with the existing datasets in terms of person categories and image number, respectively, for both the whole dataset and the test set.
Figure 3: Examples in the test set of SYSU-30k. Each pair represents a pair of images belonging to the same person category, but taken by different cameras. Left: query images; Right: gallery images.