Skip to main content

Table 6 Comparison of network accuracy (mAP and rank-1) and complexity (memory: M (parameters) and time: T (GFLOPS)) with different pruning scenarios on all the person ReID datasets

From: Exploiting prunability for person re-identification

   Market-1501 DukeMTMC-reID CUHK03-NP
Methods Scenario mAP rank-1 T M mAP rank-1 T M mAP rank-1 T M
L1 1 67.04 84.71 2.96 11.90 57.51 75.00 2.96 11.90 44.08 46.50 2.96 11.90
Taylor   66.35 84.44 3.21 12.09 57.90 75.72 3.21 12.09 44.40 46.21 3.21 12.09
Entropy   65.16 82.39 2.96 11.90 56.64 74.64 2.96 11.90 42.44 44.07 2.96 11.90
Auto-Balanced   65.46 83.64 2.96 11.90 56.45 74.64 2.96 11.90 41.85 44.21 2.96 11.90
PSFP   65.92 83.72 2.96 11.90 56.96 74.66 2.96 11.90 42.38 45.58 2.96 11.90
L1 2 49.18 70.67 2.09 8.95 40.91 61.67 2.09 8.95 26.29 27.36 2.09 8.95
Taylor   32.03 53.92 2.11 8.31 23.54 40.44 2.02 7.99 13.61 13.14 2.05 8.09
Entropy   7.38 17.31 2.09 8.95 2.34 6.78 2.09 8.95 7.83 7.29 2.09 8.95
Auto-Balanced   47.36 68.74 2.09 8.95 43.19 64.14 2.09 8.95 26.69 27.36 2.09 8.95
PSFP   65.03 80.85 2.09 8.95 53.05 73.20 2.09 8.95 42.19 43.86 2.09 8.95
L1 3 67.44 84.23 5.08 19.17 57.16 74.28 5.08 19.17 44.73 47.86 5.08 19.17
Taylor   63.29 81.38 5.28 19.60 44.14 63.33 4.98 18.53 36.91 38.86 4.96 18.46
Entropy   60.27 79.99 5.08 19.17 52.62 71.72 5.08 19.17 38.5 40.14 5.08 19.17
Auto-Balanced   67.49 84.26 5.08 19.17 58.14 75.27 5.08 19.17 46.54 48.07 5.08 19.17
PSFP   67.68 84.78 5.08 19.17 57.51 74.87 5.08 19.17 46.25 48.20 5.08 19.17
L1 4 31.41 55.70 5.08 19.17 27.94 48.79 5.08 19.17 14.97 16.07 5.08 19.17
Taylor   6.39 12.89 5.14 19.10 1.18 2.29 4.80 17.89 6.28 5.71 4.93 18.35
Entropy   25.41 46.35 5.08 19.17 21.08 41.16 5.08 19.17 11.31 11.64 5.08 19.17
Auto-Balanced   59.27 78.80 5.08 19.17 49.64 67.77 5.08 19.17 36.67 38.79 5.08 19.17
PSFP   N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A