Online Learning for Fast Segmentation of Moving Objects

Abstract

Abstract. This work addresses the problem of fast, online segmentation of moving objects in video. We pose this as a discriminative online semi-supervised appearance learning task, where supervising labels are autonomously generated by a motion segmentation algorithm. The computational complexity of the approach is significantly reduced by performing learning and classification on oversegmented image regions (superpixels), rather than per pixel. In addition, we further exploit the sparse trajectories from the motion segmentation to obtain a simple model that encodes the spatial properties and location of objects at each frame. Fusing these complementary cues produces good object segmentations at very low computational cost. In contrast to previous work, the proposed approach (1) performs segmentation on-the-fly (allowing for applications where data arrives sequentially), (2) has no prior model of object types or 'objectness', and (3) operates at significantly reduced computational cost. The approach and its ability to learn, disambiguate and segment the moving objects in the scene is evaluated on a number of benchmark video sequences.

Online Adaptation of Convolutional Neural Networks for Video Object Segmentation

“…However, our method works on the pixel level instead of the bounding box level and, in order to avoid drift, we take special care to only select training examples online for which we are very certain that they are positive or negative examples. For VOS, online adaptation is less well explored; mainly classical methods like online-updated color and/or shape models [3,4,32] and online random forests [10] have been proposed.…”

Video segmentation with joint object and trajectory labeling

“…In order to be robust to outliers that may occur due to trajectory clustering errors, we map sparse trajectory points to dense shape-location priors in the pixel-trajectory compatibility potentials. An estimate of the shape, location and scale of the foreground is computed in every frame using a kernel density estimation (KDE) [18] based on the sparse foreground points output by the binary trajectory labeling [13]. The 2D spatial distribution is estimated from the sparse points labeled as foreground (background).…”

Flying spacecraft detection with the earth as the background based on superpixels clustering

“…Then the pixel-wise labeling task is solved by maximizing a posteriori with a Markov random field (MRF) prior. However, segmenting in pixel level In order to deal with this difficulty, superpixels are introduced as a pre-processing to reduce the complexity of subsequent processing [2,[14][15][16] . The term of superpixel was coined by Ren et al [17] in 2003, indicating a group of pixels similar in color or other criteria.…”