Fast Online Object Tracking and Segmentation: A Unifying Approach

CVPR 2019 Qiang WangLi ZhangLuca BertinettoWeiming HuPhilip H. S. Torr

In this paper we illustrate how to perform both visual object tracking and semi-supervised video object segmentation, in real-time, with a single simple approach. Our method, dubbed SiamMask, improves the offline training procedure of popular fully-convolutional Siamese approaches for object tracking by augmenting their loss with a binary segmentation task... (read more)

PDF Abstract CVPR 2019 PDF CVPR 2019 Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Semi-Supervised Video Object Segmentation DAVIS 2016 SiamMask Jaccard (Mean) 71.7 # 26
Jaccard (Recall) 86.8 # 22
Jaccard (Decay) 3.0 # 27
F-measure (Mean) 67.8 # 25
F-measure (Recall) 79.8 # 22
F-measure (Decay) 2.1 # 2
J&F 69.75 # 26
Semi-Supervised Video Object Segmentation DAVIS 2017 (test-dev) SiamMask J&F 43.2 # 18
Jaccard (Mean) 40.6 # 19
Jaccard (Recall) 44.5 # 18
Jaccard (Decay) 21.9 # 10
F-measure (Mean) 45.8 # 18
F-measure (Recall) 45.3 # 19
F-measure (Decay) 22.4 # 12
Semi-Supervised Video Object Segmentation DAVIS 2017 (val) SiamMask Jaccard (Mean) 54.3 # 20
Jaccard (Recall) 62.8 # 17
Jaccard (Decay) 19.3 # 13
F-measure (Mean) 58.5 # 20
F-measure (Recall) 67.5 # 18
F-measure (Decay) 20.9 # 13
J&F 56.4 # 21
Visual Object Tracking VOT2017/18 SiamMask Expected Average Overlap (EAO) 0.380 # 6
Visual Object Tracking YouTube-VOS SiamMask Jaccard (Seen) 54.3 # 4
O (Average of Measures) 52.8 # 3
Jaccard (Unseen) 45.1 # 3
F-Measure (Seen) 58.2 # 4
F-Measure (Unseen) 47.7 # 3

Methods used in the Paper


METHOD TYPE
🤖 No Methods Found Help the community by adding them if they're not listed; e.g. Deep Residual Learning for Image Recognition uses ResNet