Online Adaptation of Convolutional Neural Networks for Video Object Segmentation

28 Jun 2017 Paul Voigtlaender Bastian Leibe

We tackle the task of semi-supervised video object segmentation, i.e. segmenting the pixels belonging to an object in the video using the ground truth pixel mask for the first frame. We build on the recently introduced one-shot video object segmentation (OSVOS) approach which uses a pretrained network and fine-tunes it on the first frame... (read more)

PDF Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Semi-Supervised Video Object Segmentation DAVIS 2016 OnAVOS Jaccard (Mean) 86.1 # 6
Jaccard (Recall) 96.1 # 9
Jaccard (Decay) 5.2 # 22
F-measure (Mean) 84.9 # 11
F-measure (Recall) 89.7 # 14
F-measure (Decay) 5.8 # 9
J&F 85.5 # 8
Semi-Supervised Video Object Segmentation DAVIS 2017 (test-dev) OnAVOS J&F 52.8 # 13
Jaccard (Mean) 49.9 # 13
Jaccard (Recall) 54.3 # 13
Jaccard (Decay) 23.0 # 12
F-measure (Mean) 55.7 # 13
F-measure (Recall) 60.3 # 16
F-measure (Decay) 23.4 # 13
Semi-Supervised Video Object Segmentation DAVIS 2017 (val) OnAVOS Jaccard (Mean) 61.6 # 15
Jaccard (Recall) 67.4 # 14
Jaccard (Decay) 27.9 # 20
F-measure (Mean) 69.1 # 13
F-measure (Recall) 75.4 # 14
F-measure (Decay) 26.6 # 18
J&F 65.35 # 16

Results from Other Papers


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK SOURCE PAPER COMPARE
Semi-Supervised Video Object Segmentation YouTube OnAVOS mIoU 0.774 # 6
Visual Object Tracking YouTube-VOS OnAVOS Jaccard (Seen) 60.1 # 2
O (Average of Measures) 55.2 # 2
Jaccard (Unseen) 46.6 # 2
F-Measure (Seen) 62.7 # 1
F-Measure (Unseen) 51.4 # 2

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