Video Segmentation by Non-Local Consensus voting

By Alon Faktor, Michal Irani. 2014

299 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Learning Features by Watching Objects Move (2016, methodology): “We adopt the NLC approach from Faktor and Irani [12].”
Tracking Emerges by Colorizing Videos (2018, background): “Tracking without Labels: We build oﬀ pioneering work for learning to segment videos without labels [51–53].”
YouTube-VOS: A Large-Scale Video Object Segmentation Benchmark (2018, unknown): “Early methods [16,28,29,30,31] for video object segmentation often solve some spatial-temporal graph structures with hand-crafted energy terms, which are usually associated with features including appearance, boundary, motion and optical ﬂows.”
Cross Pixel Optical Flow Similarity for Self-Supervised Learning (2018, unknown): “t image, thus learning a static image representation rather than a dynamic one. Closely related to our work is the use of video segmentation by [2]. They use an o-the-shelf video segmentation method [4] to construct a foreground-background segmentation dataset in an unsupervised manner. A CNN traine”
One-Shot Video Object Segmentation (2016, methodology): “We also add the unsupervised techniques: FST [36], SAL [46], KEY [24], MSG [5], TRC [13], CVOS [48], and NLC [10].”
Learning Video Object Segmentation from Static Images (2016, methodology): “Unsupervised segmentation Another family of work does general moving object segmentation (over all parts of the image), and selects post-hoc the space-time tube that best match the annotation, e.g. NLC [14] and [16,24,48].”
A Benchmark Dataset and Evaluation Methodology for Video Object Segmentation (2016, methodology): “Dynamic background (DB) scenes, e.g. flowing water, represent a major difficulty to the class of unsupervised methods, such as NLC and SAL, which adopt distinctive motion saliency as the underlying assumption to predict the object location.”
SAIL-VOS: Semantic Amodal Instance Level Video Object Segmentation – A Synthetic Dataset and Baselines (2019, background): “A variety of methods have tackled the unsupervised ( i.e ., no object annotation available during testing) [73, 71, 27, 44, 95, 96, 48] and the semi-supervised ( i.e ., ﬁrst frame annotation available during testing) [12, 54, 49, 11, 53, 98, 45, 18, 63, 105, 17, 15, 102, 100] setting.”
FusionSeg: Learning to Combine Motion and Appearance for Fully Automatic Segmentation of Generic Objects in Videos (2017, methodology): “While our approach significantly outperforms NLC [10] on the DAVIS dataset, NLC is exceptionally strong on this dataset.”
MODNet: Motion and Appearance based Moving Object Detection Network for Autonomous Driving (2018, unknown): “NLC[5] CVOS[27] KEY[14] MSG[3] FST[20] BMM[32] MPNet[29] MPNet[29]+CRF ours ours+CRF mIoU 55.”
Instance Embedding Transfer to Unsupervised Video Object Segmentation (2018, methodology): “Due to low resolution of SegTrack-v2 and the fact that SegTrack-v2 videos can have multiple moving objects of the same class in the background, we are weaker than NLC [9] (67.”
Zero-Shot Video Object Segmentation via Attentive Graph Neural Networks (2019, methodology): “To automatically separate primary objects from the background, conventional methods typically use handcrafted features ( e.g ., color, optical ﬂow) [43, 12, 59, 20] and certain heuristic assumptions related to the foreground ( i.e ., lo-cal motion differences [43], background priors [67]).”
A Survey on Deep Learning Technique for Video Segmentation (2021, unknown): “AVOS, has drawn increasing research interest [37].”
Learning Video Object Segmentation From Unlabeled Videos (2020, methodology): “…importance of capturing local consis-Supervision Non Learning Unsupervised Learning Weakly-supervised Method TRC[10] CVOS[48] KEY[24] MSG[31] NLC[7] FST[33] Motion Masks[34] TSN[17] Ours COSEG[50] MuG also outperforms classical weakly-supervised Z-VOS method COSEG [50], and all the previous…”
Saliency-Aware Video Object Segmentation (2018, methodology): “There were previous efforts [4], [5], [6], [25], [26], [27] that presented optimization frameworks for bottom-up segmentation employing both appearance and motion cues.”
Every Pixel Counts ++: Joint Learning of Geometry and Motion with 3D Holistic Understanding (2018, methodology): “For unsupervised video segmentation, saliency estimation based on 2D optical flow is often used to discover and track the objects [75], [76], and long trajectories [77] of the moving objects based on optical flow need to be considered.”
See More, Know More: Unsupervised Video Object Segmentation With Co-Attention Siamese Networks (2019, methodology): “This notion also motivated the earlier heuristic models for UVOS [14], yet it is largely ignored by current deep learning based models.”
Content-Aware Multi-Level Guidance for Interactive Instance Segmentation (2019, methodology): “Previous works have shown that most, if not all, pixels in a superpixel belong to the same category [17, 42, 25].”
Learning Motion Patterns in Videos (2016, methodology): “Measure NLC [9] CVOS [34] TRC [10] MSG [5] KEY [20] SAL [39] FST [27] PCM [3] Ours”
Video Object Segmentation with Episodic Graph Memory Networks (2020, methodology): “ Shen, L. Van Gool Table 3. Evaluation of Z-VOS on DAVIS 16 val set[34] (x4.2), with region similarity J, boundary accuracy F and time stability T . MSG NLC CUT FST SFL LMP FSEG LVO UOVOS Method [29] [12] [18] [32] [8] [44] [16] [45] [73] J Mean " 53.3 55.1 55.2 55.8 64.7 70.0 70.7 75.9 73.9 Re”