Method DMSN (SDE) on the SBM-RGBD dataset

Contact name: Ihssane Houhou

Contact email: ihssane.houhou@univ-biskra.dz

Contact Institution: University of Biskra, Algeria / University of Technology Belfort-Montbeliard, France

Method processing time: ~4 fps using Python 3.6, Keras 2.3, Tensorflow-gpu 2.1, and Cuda 10.1 installed on a machine built with Intel i9 CPU 9th generation and GeForce RTX 2080 GPU under Windows 10 operating system

Link to the code: https://github.com/ihssanehouhou/DMSN

Reference: I. Houhou, A. Zitouni, Y. Ruichek, SE. Bekhouche, M. Kas, A. Taleb-Ahmed, RGBD deep multi-scale network for background subtraction, International Journal of Multimedia Information Retrieval, vol.11, pp.395-407, May, 2022.

Get all result files: here.

A) Sequence-by-sequence results

Name	Recall	Specificity	FPR	FNR	PWC	Precision	F-Measure
Bootstrapping
BootStrapping_ds	0.9440	0.9986	0.0014	0.0560	0.6486	0.9864	0.9647
adl24cam0	0.9817	0.9998	0.0002	0.0183	0.1327	0.9964	0.9890
bear_front	0.8499	0.9987	0.0013	0.1501	4.0502	0.9956	0.9170
fall01cam0	0.9622	0.9993	0.0007	0.0378	0.2750	0.9880	0.9749
fall20cam0	0.9150	0.9992	0.0008	0.0850	0.5107	0.9848	0.9486
ColorCamouflage
Cespatx_ds	0.9658	0.9998	0.0002	0.0342	0.0883	0.9908	0.9781
Hallway	0.9587	0.9997	0.0003	0.0413	0.2463	0.9940	0.9760
colorCam1	0.9933	0.9993	0.0007	0.0067	0.2056	0.9975	0.9954
colorCam2	0.9197	0.9991	0.0009	0.0803	3.6918	0.9988	0.9576
DepthCamouflage
DCamSeq1	0.9716	0.9997	0.0003	0.0284	0.1842	0.9953	0.9833
DCamSeq2	0.9514	0.9994	0.0006	0.0486	0.2733	0.9856	0.9682
Despatx_ds	0.9603	0.9995	0.0005	0.0397	0.4098	0.9943	0.9770
Wall	0.9959	1.0000	0.0000	0.0041	0.0376	0.9999	0.9979
IlluminationChanges
ChairBox	0.9831	0.9991	0.0009	0.0169	0.2634	0.9926	0.9878
Ls_ds	0.0000	1.0000	0.0000	0.0000	0.0032	0.0000	0.0000
TimeOfDay_ds	0.0000	0.9987	0.0013	0.0000	0.1290	0.0000	0.0000
genSeq1	0.9885	0.9989	0.0011	0.0115	0.2437	0.9922	0.9903
IntermittentMotion
Shelves	0.9538	1.0000	0.0000	0.0462	0.0936	0.9983	0.9756
Sleeping_ds	0.9439	0.9987	0.0013	0.0561	1.2286	0.9947	0.9686
abandoned1	0.9893	1.0000	0.0000	0.0107	0.0285	0.9993	0.9943
abandoned2	0.9856	0.9996	0.0004	0.0144	0.2071	0.9973	0.9914
movedBackground1	0.9915	1.0000	0.0000	0.0085	0.0164	0.9974	0.9944
movedBackground2	0.9499	0.9975	0.0025	0.0501	0.5545	0.9629	0.9564
OutOfRange
MultiPeople1	0.9604	0.9994	0.0006	0.0396	0.3667	0.9925	0.9762
MultiPeople2	0.9880	0.9998	0.0002	0.0120	0.1183	0.9975	0.9927
TopViewLab1	0.9751	0.9995	0.0005	0.0249	0.1338	0.9846	0.9798
TopViewLab2	0.9842	0.9996	0.0004	0.0158	0.0955	0.9909	0.9875
TopViewLab3	0.9507	0.9995	0.0005	0.0493	0.1549	0.9759	0.9631
Shadows
Shadows_ds	0.8504	0.9992	0.0008	0.1496	0.9985	0.9851	0.9128
fall01cam1	0.9833	0.9985	0.0015	0.0167	0.2466	0.9777	0.9805
genSeq2	0.9898	0.9994	0.0006	0.0102	0.1702	0.9949	0.9923
shadows1	0.9895	0.9998	0.0002	0.0105	0.1666	0.9986	0.9940
shadows2	0.9960	0.9998	0.0002	0.0040	0.0731	0.9987	0.9973

B) Average results for each category

Name	Recall	Specificity	FPR	FNR	PWC	Precision	F-Measure
Bootstrapping	0.9306	0.9991	0.0009	0.0694	1.1234	0.9902	0.9589
ColorCamouflage	0.9594	0.9995	0.0005	0.0406	1.0580	0.9953	0.9768
DepthCamouflage	0.9698	0.9996	0.0004	0.0302	0.2262	0.9938	0.9816
IlluminationChanges	0.4929	0.9992	0.0008	0.0071	0.1598	0.4962	0.4945
IntermittentMotion	0.9690	0.9993	0.0007	0.0310	0.3548	0.9916	0.9801
OutOfRange	0.9717	0.9996	0.0004	0.0283	0.1738	0.9883	0.9799
Shadows	0.9618	0.9993	0.0007	0.0382	0.3310	0.9910	0.9754
AVG.	0.8936	0.9994	0.0006	0.0350	0.4896	0.9209	0.9067