XMem

Reported on 96 benchmarks across 3 tasks · 1 paper · 11 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision108 results

VideoonM$^3$-VOS
Average IOU· 2022-07-14
70.4
best: 75.6 (ReVOS)
SOTA
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonMOSE
F· 2022-07-14
62
best: 75.8 (Cutie+ (base, MEGA))
SOTA
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonMOSE
J· 2022-07-14
53.3
best: 67.6 (Cutie+ (base, MEGA))
SOTA
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonMOSE
J&F· 2022-07-14
57.6
best: 77.9 (SAM2)
SOTA
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonM$^3$-VOS
Average IOU· 2022-07-14
70.4
best: 75.6 (ReVOS)
SOTA
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonMOSE
F· 2022-07-14
62
best: 75.8 (Cutie+ (base, MEGA))
SOTA
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonMOSE
J· 2022-07-14
53.3
best: 67.6 (Cutie+ (base, MEGA))
SOTA
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonMOSE
J&F· 2022-07-14
57.6
best: 77.9 (SAM2)
SOTA
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonMOSE
F· 2022-07-14
62
best: 75.8 (Cutie+ (base, MEGA))
SOTA
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonMOSE
J· 2022-07-14
53.3
best: 67.6 (Cutie+ (base, MEGA))
SOTA
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonMOSE
J&F· 2022-07-14
57.6
best: 77.9 (SAM2)
SOTA
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS-2017 (test-dev)
F-measure· 2022-07-14
84.5
best: 87 (XMem (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS-2017 (test-dev)
Jaccard· 2022-07-14
77.4
best: 80.5 (XMem (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS-2017 (test-dev)
Mean Jaccard & F-Measure· 2022-07-14
81
best: 83.7 (XMem (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2019
F-Measure (Seen)· uses extra data· 2022-07-14
88.6
best: 90.6 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2019
F-Measure (Unseen)· uses extra data· 2022-07-14
88.6
best: 90.5 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2019
Jaccard (Seen)· uses extra data· 2022-07-14
84.3
best: 86.3 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2019
Jaccard (Unseen)· uses extra data· 2022-07-14
80.3
best: 754.8 (R50-AOST (L'=1))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2019
Mean Jaccard & F-Measure· uses extra data· 2022-07-14
85.5
best: 86.8 (XMem (BL30K,MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2016
F-Score· 2022-07-14
92.7
best: 94.7 (AOC-MF (val))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2016
J&F· 2022-07-14
91.5
best: 93.4 (ISVOS (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2016
Jaccard (Mean)· 2022-07-14
90.4
best: 92.5 (ISVOS (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2017 (val)
F-measure· 2022-07-14
89.5
best: 92.6 (XMem (BLK30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2017 (val)
Jaccard· 2022-07-14
82.9
best: 86.3 (XMem (BLK30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2017 (val)
Mean Jaccard & F-Measure· 2022-07-14
86.2
best: 89.5 (XMem (BLK30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2017 (val)
F-measure (Mean)· uses extra data· 2022-07-14
89.5
best: 93.4 (Cutie+ (base))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2017 (val)
J&F· uses extra data· 2022-07-14
86.2
best: 90.7 (SAM2)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2017 (val)
Jaccard (Mean)· uses extra data· 2022-07-14
82.9
best: 87.5 (Cutie+ (base))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2017 (val)
Speed (FPS)· uses extra data· 2022-07-14
22.6
best: 90.6 (MobileVOS (BL30K))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2016
F-measure (Mean)· uses extra data· 2022-07-14
92.7
best: 94.7 (SwinB-DeAOT-L)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2016
J&F· uses extra data· 2022-07-14
91.5
best: 93.4 (ISVOS (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2016
Jaccard (Mean)· uses extra data· 2022-07-14
90.4
best: 92.5 (ISVOS (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2016
Speed (FPS)· uses extra data· 2022-07-14
29.6
best: 100.1 (MobileVOS (BL30K))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2019
F-Measure (Seen)· 2022-07-14
88
best: 90.6 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2019
F-Measure (Unseen)· 2022-07-14
87.1
best: 90.5 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2019
Jaccard (Seen)· 2022-07-14
83.6
best: 86.3 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2019
Jaccard (Unseen)· 2022-07-14
78.5
best: 754.8 (R50-AOST (L'=1))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2019
Overall· 2022-07-14
84.3
best: 87.5 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2017 (test-dev)
F-measure (Mean)· uses extra data· 2022-07-14
84.5
best: 91.4 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2017 (test-dev)
J&F· uses extra data· 2022-07-14
81
best: 88.1 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS 2017 (test-dev)
Jaccard (Mean)· uses extra data· 2022-07-14
77.4
best: 84.7 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonDAVIS (no YouTube-VOS training)
FPS· 2022-07-14
29.6
best: 50.1 (TBD)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2018
F-Measure (Seen)· uses extra data· 2022-07-14
89.3
best: 91 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2018
F-Measure (Unseen)· uses extra data· 2022-07-14
88.7
best: 90.2 (XMem (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2018
Jaccard (Seen)· uses extra data· 2022-07-14
84.6
best: 86.6 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2018
Jaccard (Unseen)· uses extra data· 2022-07-14
80.2
best: 82.2 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2018
Overall· uses extra data· 2022-07-14
85.7
best: 87.5 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
VideoonYouTube-VOS 2018
Speed (FPS)· uses extra data· 2022-07-14
22.6
best: 65.9 (FRTM)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS-2017 (test-dev)
F-measure· 2022-07-14
84.5
best: 87 (XMem (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS-2017 (test-dev)
Jaccard· 2022-07-14
77.4
best: 80.5 (XMem (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS-2017 (test-dev)
Mean Jaccard & F-Measure· 2022-07-14
81
best: 83.7 (XMem (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2019
F-Measure (Seen)· uses extra data· 2022-07-14
88.6
best: 90.6 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2019
F-Measure (Unseen)· uses extra data· 2022-07-14
88.6
best: 90.5 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2019
Jaccard (Seen)· uses extra data· 2022-07-14
84.3
best: 86.3 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2019
Jaccard (Unseen)· uses extra data· 2022-07-14
80.3
best: 754.8 (R50-AOST (L'=1))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2019
Mean Jaccard & F-Measure· uses extra data· 2022-07-14
85.5
best: 86.8 (XMem (BL30K,MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2016
F-Score· 2022-07-14
92.7
best: 94.7 (AOC-MF (val))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2016
J&F· 2022-07-14
91.5
best: 93.4 (ISVOS (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2016
Jaccard (Mean)· 2022-07-14
90.4
best: 92.5 (ISVOS (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2017 (val)
F-measure· 2022-07-14
89.5
best: 92.6 (XMem (BLK30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2017 (val)
Jaccard· 2022-07-14
82.9
best: 86.3 (XMem (BLK30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2017 (val)
Mean Jaccard & F-Measure· 2022-07-14
86.2
best: 89.5 (XMem (BLK30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2017 (val)
F-measure (Mean)· uses extra data· 2022-07-14
89.5
best: 93.4 (Cutie+ (base))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2017 (val)
J&F· uses extra data· 2022-07-14
86.2
best: 90.7 (SAM2)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2017 (val)
Jaccard (Mean)· uses extra data· 2022-07-14
82.9
best: 87.5 (Cutie+ (base))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2017 (val)
Speed (FPS)· uses extra data· 2022-07-14
22.6
best: 90.6 (MobileVOS (BL30K))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2016
F-measure (Mean)· uses extra data· 2022-07-14
92.7
best: 94.7 (SwinB-DeAOT-L)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2016
J&F· uses extra data· 2022-07-14
91.5
best: 93.4 (ISVOS (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2016
Jaccard (Mean)· uses extra data· 2022-07-14
90.4
best: 92.5 (ISVOS (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2016
Speed (FPS)· uses extra data· 2022-07-14
29.6
best: 100.1 (MobileVOS (BL30K))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2019
F-Measure (Seen)· 2022-07-14
88
best: 90.6 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2019
F-Measure (Unseen)· 2022-07-14
87.1
best: 90.5 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2019
Jaccard (Seen)· 2022-07-14
83.6
best: 86.3 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2019
Jaccard (Unseen)· 2022-07-14
78.5
best: 754.8 (R50-AOST (L'=1))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2019
Overall· 2022-07-14
84.3
best: 87.5 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2017 (test-dev)
F-measure (Mean)· uses extra data· 2022-07-14
84.5
best: 91.4 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2017 (test-dev)
J&F· uses extra data· 2022-07-14
81
best: 88.1 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS 2017 (test-dev)
Jaccard (Mean)· uses extra data· 2022-07-14
77.4
best: 84.7 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonDAVIS (no YouTube-VOS training)
FPS· 2022-07-14
29.6
best: 50.1 (TBD)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2018
F-Measure (Seen)· uses extra data· 2022-07-14
89.3
best: 91 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2018
F-Measure (Unseen)· uses extra data· 2022-07-14
88.7
best: 90.2 (XMem (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2018
Jaccard (Seen)· uses extra data· 2022-07-14
84.6
best: 86.6 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2018
Jaccard (Unseen)· uses extra data· 2022-07-14
80.2
best: 82.2 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2018
Overall· uses extra data· 2022-07-14
85.7
best: 87.5 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Video Object SegmentationonYouTube-VOS 2018
Speed (FPS)· uses extra data· 2022-07-14
22.6
best: 65.9 (FRTM)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonDAVIS 2017 (val)
F-measure (Mean)· uses extra data· 2022-07-14
89.5
best: 93.4 (Cutie+ (base))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonDAVIS 2017 (val)
J&F· uses extra data· 2022-07-14
86.2
best: 90.7 (SAM2)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonDAVIS 2017 (val)
Jaccard (Mean)· uses extra data· 2022-07-14
82.9
best: 87.5 (Cutie+ (base))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonDAVIS 2017 (val)
Speed (FPS)· uses extra data· 2022-07-14
22.6
best: 90.6 (MobileVOS (BL30K))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonDAVIS 2016
F-measure (Mean)· uses extra data· 2022-07-14
92.7
best: 94.7 (SwinB-DeAOT-L)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonDAVIS 2016
J&F· uses extra data· 2022-07-14
91.5
best: 93.4 (ISVOS (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonDAVIS 2016
Jaccard (Mean)· uses extra data· 2022-07-14
90.4
best: 92.5 (ISVOS (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonDAVIS 2016
Speed (FPS)· uses extra data· 2022-07-14
29.6
best: 100.1 (MobileVOS (BL30K))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonYouTube-VOS 2019
F-Measure (Seen)· 2022-07-14
88
best: 90.6 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonYouTube-VOS 2019
F-Measure (Unseen)· 2022-07-14
87.1
best: 90.5 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonYouTube-VOS 2019
Jaccard (Seen)· 2022-07-14
83.6
best: 86.3 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonYouTube-VOS 2019
Jaccard (Unseen)· 2022-07-14
78.5
best: 754.8 (R50-AOST (L'=1))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonYouTube-VOS 2019
Overall· 2022-07-14
84.3
best: 87.5 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonDAVIS 2017 (test-dev)
F-measure (Mean)· uses extra data· 2022-07-14
84.5
best: 91.4 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonDAVIS 2017 (test-dev)
J&F· uses extra data· 2022-07-14
81
best: 88.1 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonDAVIS 2017 (test-dev)
Jaccard (Mean)· uses extra data· 2022-07-14
77.4
best: 84.7 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonDAVIS (no YouTube-VOS training)
FPS· 2022-07-14
29.6
best: 50.1 (TBD)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonYouTube-VOS 2018
F-Measure (Seen)· uses extra data· 2022-07-14
89.3
best: 91 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonYouTube-VOS 2018
F-Measure (Unseen)· uses extra data· 2022-07-14
88.7
best: 90.2 (XMem (BL30K, MS))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonYouTube-VOS 2018
Jaccard (Seen)· uses extra data· 2022-07-14
84.6
best: 86.6 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonYouTube-VOS 2018
Jaccard (Unseen)· uses extra data· 2022-07-14
80.2
best: 82.2 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonYouTube-VOS 2018
Overall· uses extra data· 2022-07-14
85.7
best: 87.5 (Cutie+ (base, MEGA))
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115
Semi-Supervised Video Object SegmentationonYouTube-VOS 2018
Speed (FPS)· uses extra data· 2022-07-14
22.6
best: 65.9 (FRTM)
XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model arXiv:2207.07115