Fig. 1

A brief chronology of crowd counting. The first incorporation of deep learning techniques for crowd counting is from 2015. See Section 1 for more detailed description. Milestone models in this figure: MLR [49], KRR [50], Chan et al. [39], Lemptisky et al. [16], RR [41], CA-RR [51], Count Forest [52], Wang et al. [53], Fu et al. [54], Cross scene [55], MCNN [1], Hydra-CNN [2], CP-CNN [6], CMTL [56], switching CNN [5], CSRNet [12], SANet [11], PSSDN [57], LSF-CNN [58], JHU-CROWD [59], NWPU-Crowd [28], Auto-Scale [60], GLoss [61], CLTR [62], CSS-CCNN [63], ZSC [64], PET [65], mPrompt [66] and CrowdDiff [67]. Traditional heuristic methods are highlighted with the blue-shaded area and the modern CNN-based density estimation and crowd counting models are with the red-shaded backgrounds, respectively