Parallel Correlation Filter Tracking Algorithm Based on Response Map Confidence

Qiqi Song; Xiaoli Li; Wei Zuo; Lipeng Gu

doi:10.3788/LOP57.201017

Journals >Laser & Optoelectronics Progress >Volume 57 >Issue 20 >Page 201017 > Article

Laser & Optoelectronics Progress
Vol. 57, Issue 20, 201017 (2020)

Parallel Correlation Filter Tracking Algorithm Based on Response Map Confidence

Qiqi Song, Xiaoli Li^*, Wei Zuo, and Lipeng Gu

Author Affiliations

School of Information Science and Technology, Donghua University, Shanghai 201620, China

show less

DOI: 10.3788/LOP57.201017 Cite this Article Set citation alerts

Qiqi Song, Xiaoli Li, Wei Zuo, Lipeng Gu. Parallel Correlation Filter Tracking Algorithm Based on Response Map Confidence[J]. Laser & Optoelectronics Progress, 2020, 57(20): 201017 Copy Citation Text

show less

Fig. 1. Result when the trace fails. (a1)-(c1) Images of different frames; (a2)-(c2) corresponding response images

Download full size

Fig. 2. Change curves of the three indicators of the Human4 sequence

Download full size

Fig. 3. OPE of different algorithms on the OTB-2015 dataset. (a) Success rate; (b) precision

Download full size

Fig. 4. Tracking effect of different algorithms on the OTB-2015 dataset. (a) Coke; (b) bolt

Download full size

Fig. 5. Test results of different algorithms. (a) OPE success rate (OTB-2013 dataset); (b) OPE precision (OTB-2013dataset); (c) OPE success rate (OTB-2015 dataset); (d) OPE precision (OTB-2015) dataset)

Download full size

Fig. 6. Tracking results of different algorithms. (a) BlurOwl; (b) Diving; (c) Walking2; (d) Couple

Download full size

Fig. 7. Case of target tracking failure. (a) Bird1; (b) Biker

Download full size

Tracker	SV	IV	OPR	OCC	BC	DEF	MB	FM	IPR	OV	LR	AUC
Ours	61.1	73.4	64.2	62.3	72.7	59.5	65.9	64.5	67.2	59.5	61.9	68.7
SAMF	58.7	64.0	66.2	66.7	63.9	60.9	64.1	59.8	64.1	55.1	51.5	67.3
Staple	58.9	70.4	63.4	63.3	65.4	62.5	61.2	60.8	64.0	54.1	43.2	67.2
FDSST	59.8	72.0	59.5	59.4	72.8	57.3	66.4	67.1	66.1	54.8	58.9	66.3
LMCF	57.4	66.7	60.1	61.0	67.1	58.3	64.7	58.8	58.2	58.0	51.2	64.7
LCT	54.3	64.1	61.3	55.8	65.5	56.9	57.5	57.2	61.0	47.5	36.9	63.7
DSST	52.2	57.5	50.3	49.9	57.4	44.3	51.6	51.9	55.1	47.8	44.3	57.0
KCF	37.3	46.4	44.4	41.4	52.9	43.7	47.6	47.8	49.0	39.3	28.4	48.7
CSK	30.9	39.9	37.8	35.3	46.5	35.7	34.5	36.8	41.3	27.4	27.7	41.1
DFT	27.0	37.8	36.9	37.1	43.1	38.2	28.3	29.3	36.2	31.3	19.6	35.8

Table 1. Success rate of 10 algorithms in the OTB-2015 dataset unit: %

Tracker	SV	IV	OPR	OCC	BC	DEF	MB	FM	IPR	OV	LR	AP
Ours	71.6	78.8	73.3	68.4	79.4	66.7	69.1	69.5	77.1	64.8	68.5	76.1
SAMF	70.2	70.8	73.9	72.2	68.9	68.0	65.5	65.5	72.1	62.8	68.5	75.1
Staple	72.5	77.8	73.0	70.9	71.4	70.5	63.9	66.6	75.3	59.3	60.9	75.1
FDSST	67.2	76.5	66.6	64.6	78.0	62.0	69.1	69.8	73.4	57.7	64.1	72.5
LMCF	67.7	73.8	68.7	68.2	72.0	65.2	66.5	63.5	67.5	57.2	67.9	71.9
LCT	63.0	69.7	69.0	60.3	66.3	64.8	54.9	57.2	70.7	43.6	50.5	69.9
DSST	62.5	63.2	57.1	54.4	62.2	50.3	51.4	51.9	64.1	50.1	64.3	63.3
KCF	58.0	63.2	59.5	52.6	62.2	55.9	50.5	54.8	63.4	37.1	54.6	62.2
CSK	45.2	49.0	48.9	43.3	57.4	45.7	35.5	39.7	51.4	27.6	42.2	51.8
DFT	35.5	41.6	44.0	42.0	45.5	41.7	27.2	30.5	43.4	35.3	29.0	41.4

Table 2. Precision of 10 algorithms on the OTB-2015 dataset unit:%

Qiqi Song, Xiaoli Li, Wei Zuo, Lipeng Gu. Parallel Correlation Filter Tracking Algorithm Based on Response Map Confidence[J]. Laser & Optoelectronics Progress, 2020, 57(20): 201017

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information