Method AvgM-D on the SBM-RGBD dataset
Contact name: Guangli Li^1,2 and Xueying Wang^2
Contact email: liguangli@ict.ac.cn and wangxueying@ict.ac.cn
Contact Institution: ^1 ICT, Chinese Academy of Sciences (Beijing, China) and ^2 Jilin University (Changchun, China)
Method processing time: From ~2 to ~9 fps for 640x480 videos (python code running on a Xeon E5 server)
Method parameters: lambda=0.0001, alpha_h=9, alpha_w=9
Note: Uses only depth data
Reference: Unpublished
Get all result files:
here.
A) Sequence-by-sequence results
Name | Recall | Specificity | FPR | FNR | PWC | Precision | F-Measure |
Bootstrapping |
BootStrapping_ds | 0.8733 | 0.9910 | 0.0090 | 0.1267 | 2.0028 | 0.9098 | 0.8912 |
adl24cam0 | 0.1678 | 0.9818 | 0.0182 | 0.8322 | 6.7601 | 0.3729 | 0.2315 |
bear_front | 0.3246 | 0.9861 | 0.0139 | 0.6754 | 18.8037 | 0.8926 | 0.4760 |
fall01cam0 | 0.5334 | 0.9835 | 0.0165 | 0.4666 | 4.1483 | 0.6561 | 0.5884 |
fall20cam0 | 0.3945 | 0.9879 | 0.0121 | 0.6055 | 4.2648 | 0.6394 | 0.4880 |
ColorCamouflage |
Cespatx_ds | 0.9107 | 0.9942 | 0.0058 | 0.0893 | 0.7558 | 0.7651 | 0.8316 |
Hallway | 0.6969 | 0.9681 | 0.0319 | 0.3031 | 4.6098 | 0.5464 | 0.6125 |
colorCam1 | 0.9959 | 0.9905 | 0.0095 | 0.0041 | 0.8262 | 0.9681 | 0.9818 |
colorCam2 | 0.9967 | 0.9643 | 0.0357 | 0.0033 | 2.0958 | 0.9587 | 0.9774 |
DepthCamouflage |
DCamSeq1 | 0.9557 | 0.9906 | 0.0094 | 0.0443 | 1.1300 | 0.8561 | 0.9032 |
DCamSeq2 | 0.5965 | 0.9960 | 0.0040 | 0.4035 | 2.1408 | 0.8710 | 0.7081 |
Despatx_ds | 0.8154 | 0.9894 | 0.0106 | 0.1846 | 2.6637 | 0.8871 | 0.8497 |
Wall | 0.9796 | 0.9927 | 0.0073 | 0.0204 | 0.8428 | 0.9299 | 0.9541 |
IlluminationChanges |
ChairBox | 0.3944 | 0.9914 | 0.0086 | 0.6056 | 7.3596 | 0.8494 | 0.5386 |
Ls_ds | 0.0000 | 0.9917 | 0.0083 | 0.0000 | 0.8259 | 0.0000 | 0.0000 |
TimeOfDay_ds | 0.0000 | 0.9893 | 0.0107 | 0.0000 | 1.0663 | 0.0000 | 0.0000 |
genSeq1 | 0.9623 | 0.9707 | 0.0293 | 0.0377 | 3.0350 | 0.8257 | 0.8888 |
IntermittentMotion |
Shelves | 0.7496 | 0.9986 | 0.0014 | 0.2504 | 0.6249 | 0.9160 | 0.8245 |
Sleeping_ds | 0.8651 | 0.9716 | 0.0284 | 0.1349 | 4.9977 | 0.8852 | 0.8751 |
abandoned1 | 0.9745 | 0.9977 | 0.0023 | 0.0255 | 0.2915 | 0.9146 | 0.9436 |
abandoned2 | 0.9865 | 0.9875 | 0.0125 | 0.0135 | 1.2633 | 0.9159 | 0.9499 |
movedBackground1 | 0.9225 | 0.9992 | 0.0008 | 0.0775 | 0.1979 | 0.9424 | 0.9323 |
movedBackground2 | 0.8874 | 0.9929 | 0.0071 | 0.1126 | 1.3867 | 0.8949 | 0.8912 |
OutOfRange |
MultiPeople1 | 0.8083 | 0.9870 | 0.0130 | 0.1917 | 2.6935 | 0.8411 | 0.8243 |
MultiPeople2 | 0.8915 | 0.9956 | 0.0044 | 0.1085 | 1.2912 | 0.9470 | 0.9184 |
TopViewLab1 | 0.5136 | 0.9845 | 0.0155 | 0.4864 | 3.1168 | 0.5333 | 0.5233 |
TopViewLab2 | 0.4987 | 0.9806 | 0.0194 | 0.5013 | 3.7913 | 0.5077 | 0.5031 |
TopViewLab3 | 0.4474 | 0.9820 | 0.0180 | 0.5526 | 2.9387 | 0.3508 | 0.3933 |
Shadows |
Shadows_ds | 0.9536 | 0.9946 | 0.0054 | 0.0464 | 0.7893 | 0.9190 | 0.9360 |
fall01cam1 | 0.5328 | 0.9939 | 0.0061 | 0.4672 | 3.5140 | 0.8546 | 0.6564 |
genSeq2 | 0.9517 | 0.9865 | 0.0135 | 0.0483 | 1.7337 | 0.8988 | 0.9245 |
shadows1 | 0.9813 | 0.9823 | 0.0177 | 0.0187 | 1.7880 | 0.9002 | 0.9390 |
shadows2 | 0.9867 | 0.9808 | 0.0192 | 0.0133 | 1.8399 | 0.8907 | 0.9363 |
B) Average results for each category
Name | Recall | Specificity | FPR | FNR | PWC | Precision | F-Measure |
Bootstrapping | 0.4587 | 0.9861 | 0.0139 | 0.5413 | 7.1960 | 0.6941 | 0.5350 |
ColorCamouflage | 0.9001 | 0.9793 | 0.0207 | 0.0999 | 2.0719 | 0.8096 | 0.8508 |
DepthCamouflage | 0.8368 | 0.9922 | 0.0078 | 0.1632 | 1.6943 | 0.8860 | 0.8538 |
IlluminationChanges | 0.3392 | 0.9858 | 0.0142 | 0.1608 | 3.0717 | 0.4188 | 0.3569 |
IntermittentMotion | 0.8976 | 0.9912 | 0.0088 | 0.1024 | 1.4603 | 0.9115 | 0.9027 |
OutOfRange | 0.6319 | 0.9860 | 0.0140 | 0.3681 | 2.7663 | 0.6360 | 0.6325 |
Shadows | 0.8812 | 0.9876 | 0.0124 | 0.1188 | 1.9330 | 0.8927 | 0.8784 |
AVG. | 0.7065 | 0.9869 | 0.0131 | 0.2221 | 2.8848 | 0.7498 | 0.7157 |