Method MFCN on the SBM-RGBD dataset

Contact name: Dengdong Zeng

Contact email: zengdongdong13@mails.ucas.edu.cn

Contact Institution: University of Chinese Academy of Sciences

Method processing time: ~27 fps for all videos with TGX 1060 GPU

Reference: D. Zeng and M. Zhu, Background Subtraction Using Multiscale Fully Convolutional Network, IEEE Access, 6, 2018.

Get all result files: here.

A) Sequence-by-sequence results

Name	Recall	Specificity	FPR	FNR	PWC	Precision	F-Measure
Bootstrapping
BootStrapping_ds	0.9824	0.9981	0.0019	0.0176	0.3332	0.9822	0.9823
adl24cam0	0.9927	0.9995	0.0005	0.0073	0.0927	0.9920	0.9924
bear_front	0.9958	0.9960	0.0040	0.0042	0.4072	0.9888	0.9923
fall01cam0	0.9796	0.9994	0.0006	0.0204	0.1730	0.9891	0.9844
fall20cam0	0.9826	0.9995	0.0005	0.0174	0.1371	0.9907	0.9866
ColorCamouflage
Cespatx_ds	0.9774	0.9997	0.0003	0.0226	0.0750	0.9857	0.9816
Hallway	0.9899	0.9994	0.0006	0.0101	0.1143	0.9882	0.9891
colorCam1	0.9965	0.9969	0.0031	0.0035	0.3182	0.9893	0.9929
colorCam2	0.9797	0.9949	0.0051	0.0203	1.2011	0.9937	0.9867
DepthCamouflage
DCamSeq1	0.9829	0.9970	0.0030	0.0171	0.3804	0.9506	0.9665
DCamSeq2	0.9774	0.9976	0.0024	0.0226	0.3244	0.9499	0.9635
Despatx_ds	0.9884	0.9996	0.0004	0.0116	0.1377	0.9964	0.9924
Wall	0.9992	1.0000	0.0000	0.0008	0.0112	0.9995	0.9994
IlluminationChanges
ChairBox	0.9963	0.9989	0.0011	0.0037	0.1348	0.9914	0.9938
Ls_ds	0.0000	1.0000	0.0000	0.0000	0.0000	0.0000	0.0000
TimeOfDay_ds	0.0000	1.0000	0.0000	0.0000	0.0000	0.0000	0.0000
genSeq1	0.9983	0.9960	0.0040	0.0017	0.3672	0.9732	0.9856
IntermittentMotion
Shelves	0.9899	0.9999	0.0001	0.0101	0.0272	0.9961	0.9930
Sleeping_ds	0.9657	0.9971	0.0029	0.0343	0.9245	0.9881	0.9767
abandoned1	0.9982	0.9998	0.0002	0.0018	0.0262	0.9914	0.9948
abandoned2	0.9973	0.9991	0.0009	0.0027	0.1080	0.9938	0.9956
movedBackground1	0.9974	0.9997	0.0003	0.0026	0.0314	0.9817	0.9895
movedBackground2	0.9954	0.9964	0.0036	0.0046	0.3625	0.9503	0.9723
OutOfRange
MultiPeople1	0.9948	0.9979	0.0021	0.0052	0.2358	0.9755	0.9851
MultiPeople2	0.9968	0.9974	0.0026	0.0032	0.2685	0.9710	0.9838
TopViewLab1	0.9919	0.9985	0.0015	0.0081	0.1693	0.9587	0.9750
TopViewLab2	0.9957	0.9987	0.0013	0.0043	0.1416	0.9684	0.9819
TopViewLab3	0.9794	0.9985	0.0015	0.0206	0.1938	0.9328	0.9556
Shadows
Shadows_ds	0.9518	0.9991	0.0009	0.0482	0.3846	0.9852	0.9682
fall01cam1	0.9977	0.9983	0.0017	0.0023	0.1729	0.9754	0.9864
genSeq2	0.9984	0.9974	0.0026	0.0016	0.2532	0.9793	0.9888
shadows1	0.9994	0.9987	0.0013	0.0006	0.1179	0.9922	0.9958
shadows2	0.9995	0.9982	0.0018	0.0005	0.1603	0.9889	0.9942

B) Average results for each category

Name	Recall	Specificity	FPR	FNR	PWC	Precision	F-Measure
Bootstrapping	0.9866	0.9985	0.0015	0.0134	0.2286	0.9885	0.9876
ColorCamouflage	0.9859	0.9977	0.0023	0.0141	0.4272	0.9893	0.9876
DepthCamouflage	0.9870	0.9986	0.0014	0.0130	0.2134	0.9741	0.9804
IlluminationChanges	0.4986	0.9987	0.0013	0.0014	0.1255	0.4912	0.4949
IntermittentMotion	0.9906	0.9987	0.0013	0.0094	0.2466	0.9836	0.9870
OutOfRange	0.9917	0.9982	0.0018	0.0083	0.2018	0.9613	0.9763
Shadows	0.9893	0.9983	0.0017	0.0107	0.2178	0.9842	0.9867
AVG.	0.9186	0.9984	0.0016	0.0100	0.2373	0.9103	0.9143