Lucid Data Dreaming for Object Tracking

By Anna Khoreva, Rodrigo Benenson, Eddy Ilg, Thomas Brox, Bernt Schiele. 2017

139 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

Iteratively Trained Interactive Segmentation (2018, background): “Many recent works [4, 15, 23, 26] focus on segmenting objects in videos since such object annotations are expensive.”
Meta-Transfer Learning for Few-Shot Learning (2018, neutral): “Data augmentation is a classic technique to increase the amount of available data and thus also useful for few-shot learning [21].”
Meta-Transfer Learning Through Hard Tasks (2019, background): “amount of available data by data augmentation techniques [10].”
Video Object Segmentation without Temporal Information (2017, background): “performingmethods of the challenge [29], [33], [35], [61].”
Interactive Video Object Segmentation in the Wild (2017, background): “Finally, the user can optionally reﬁne poorly segmented frames with 1-2 clicks. mentation masks of all frames in the sequence [3, 9, 15].”
A Generative Appearance Model for End-To-End Video Object Segmentation (2018, background): “The approach was extended in a number of works: continuous training during the sequence [30]; adding instance-level semantic information [21]; incorporating motion information via optical flow [1, 6, 12]; performing temporal propagation via a Markov random field [1]; location-specific embeddings [8]”
FEELVOS: Fast End-To-End Embedding Learning for Video Object Segmentation (2019, methodology): “This approach is extended by LucidTracker [20] which introduces an elaborate data augmentation mechanism.”
Video Object Segmentation with Joint Re-identification and Attention-Aware Mask Propagation (2018, methodology): “For online training, we follow [16] to synthesize videos based on the first frame of testing videos and add them into the training set.”
VideoMatch: Matching based Video Object Segmentation (2018, background): “OURS OURS-NU OSVOS− MSK− OnAVOS− SFL− VPN CTN mIoU 0.810 0.792 0.525 0.699 0.736 0.674 0.702 0.735 Speed (s) 0.32 0.17 0.12 0.15 3.55 0.3 0.63 29.95 10 -1 10 0 10 1 10 2 10 3 Time per frame (s) 0.5 0.6 0.7 0.8 0.9 1 m Io U BVS HVS SEA JMP OFL OSVOS OnAVOS VPN SFL MSK PLM Lucid OURS-NU OURS (a) mIoU ”
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.”
Video Object Segmentation Using Space-Time Memory Networks (2019, methodology): “To make the network object-specific, online training data is generated from the first frame by deforming the object mask [26] or synthesizing images [14] for fine-tuning.”
Online Adaptation of Convolutional Neural Networks for Video Object Segmentation (2017, methodology): “Current state of the art methods use post-processing steps such as boundary snapping [7], or conditional random ﬁeld (CRF) smoothing [24, 35] to improve the contours.”
Fast Video Object Segmentation by Reference-Guided Mask Propagation (2018, methodology): “Some methods are also aided by additional optical flow information [30, 21].”
See More, Know More: Unsupervised Video Object Segmentation With Co-Attention Siamese Networks (2019, methodology): “In addition, both object tracking [29, 8, 12, 36] and person re-identification [34, 66] have been fused into SVOS task to handle deformation and occlusion issues.”
Fast Online Object Tracking and Segmentation: A Unifying Approach (2018, methodology): “This result is achieved with a simple bounding box initialisation (as opposed to a mask) and without adopting costly techniques often used by VOS approaches such as fine-tuning [39, 45, 2, 60], data augmentation [25, 33] and optical flow [57, 2, 45, 33, 10].”