Previous deep learning-based video stabilizers require a large scale of
paired unstable and stable videos for training, which are difficult to collect.
Traditional trajectory-based stabilizers, on the other hand, divide the task
into several sub-tasks and tackle them subsequently, which are fragile in
textureless and occluded regions regarding the usage of hand-crafted features.
In this paper, we attempt to tackle the video stabilization problem in a deep
unsupervised learning manner, which borrows the divide-and-conquer idea from
traditional stabilizers while leveraging the representation power of DNNs to
handle the challenges in real-world scenarios. Technically, DUT is composed of
a trajectory estimation stage and a trajectory smoothing stage. In the
trajectory estimation stage, we first estimate the motion of keypoints,
initialize and refine the motion of grids via a novel multi-homography
estimation strategy and a motion refinement network, respectively, and get the
grid-based trajectories via temporal association. In the trajectory smoothing
stage, we devise a novel network to predict dynamic smoothing kernels for
trajectory smoothing, which can well adapt to trajectories with different
dynamic patterns. We exploit the spatial and temporal coherence of keypoints
and grid vertices to formulate the training objectives, resulting in an
unsupervised training scheme. Experiment results on public benchmarks show that
DUT outperforms state-of-the-art methods both qualitatively and quantitatively.
The source code is available at this https URL

本文提出了一种基于深度非监督学习的视频稳定方法 DUT，该方法通过将传统稳定器的任务分解为多个子任务，并借助 DNNs 的表征能力来处理现实世界场景中的挑战，以及利用关键点和网格顶点的时空一致性来制定训练目标，从而实现了无监督训练，并在公共基准测试中表现优异。

通过观看不稳定视频学习视频稳定技术

DUT: Learning Video Stabilization by Simply Watching Unstable Videos

In the recent past, several sampling-based algorithms have been proposed to
compute trajectories that are collision-free and dynamically-feasible. However,
the outputs of such algorithms are notoriously jagged. In this paper, by
focusing on robots with car-like dynamics, we present a fast and simple
heuristic algorithm, named Convex Elastic Smoothing (CES) algorithm, for
trajectory smoothing and speed optimization. The CES algorithm is inspired by
earlier work on elastic band planning and iteratively performs shape and speed
optimization. The key feature of the algorithm is that both optimization
problems can be solved via convex programming, making CES particularly fast. A
range of numerical experiments show that the CES algorithm returns high-quality
solutions in a matter of a few hundreds of milliseconds and hence appears
amenable to a real-time implementation.

本文提出 CES 算法，解决基于采样的算法在计算碰撞避免和动态可行性轨迹后输出曲线过于锯齿状的问题，该算法主要应用于类似于汽车动力学的机器人，采用了基于凸规划的形状和速度迭代优化，取得了很好的效果，几百毫秒内就能够产生高质量的轨迹曲线设计，适合实时实现。