MOT Challenge

Video not available.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Benchmark:

CVPR 2020 MOTS Challenge |

Short name:

DD_Vision

Description:

This tracking method is an adaptation of the previous work UnOVOST.

Reference:

L. Yang Liu. Tracking by Segmentation: Person-ReID and Optical Flow Based Offline Tracker for the MOTSChallenge 2020. In , 2020.

Last submitted:

June 08, 2020 (3 years ago)

Published:

September 30, 2020 at 16:22:54 CET

Submissions:

Project page / code:

n/a

Open source:

Hardware:

Tesla P40

Runtime:

1.6 Hz

Benchmark performance:

Sequence	sMOTSA	IDF1	MOTSA	MOTSP	MODSA	MT	ML	TP	FP	FN	Rcll	Prcn	ID Sw.	Frag
CVPR 2020 MOTS Challenge	66.6	71.8	79.7	84.4	80.7	243 (74.1)	15 (4.6)	27,114	1,067	5,155	84.0	96.2	341 (405.8)	559 (665.3)

Detailed performance:

Sequence	sMOTSA	IDF1	MOTSA	MOTSP	MODSA	MT	ML	TP	FP	FN	Rcll	Prcn	ID Sw.	Frag
MOTS20-01	68.5	84.2	84.3	82.3	85.3	9	0	2,781	131	325	89.5	95.5	31	51
MOTS20-06	75.3	66.4	89.2	85.2	90.5	163	2	9,216	333	598	93.9	96.5	129	214
MOTS20-07	58.0	75.7	70.6	82.7	71.6	21	11	9,317	94	3,561	72.3	99.0	136	212
MOTS20-12	69.4	67.1	81.1	87.0	81.8	50	2	5,800	509	671	89.6	91.9	45	82

Raw data:

download

GMPHD_SAF

DD_Vision

TrackRCNN

DD_Vision: Tracking by Segmentation: Person-ReID and Optical Flow Based Offline Tracker