Optical flow estimation is a classical yet challenging task in computer vision.

Optical flow estimation is a classical yet challenging task in computer vision. One of the essential factors in accurately predicting optical flow is to alleviate occlusions between frames. However, it is still a thorny problem for current top-performing optical flow estimation methods due to insufficient local evidence to model occluded areas. In this paper, we propose the Super Kernel Flow Network (SKFlow), a CNN architecture to ameliorate the impacts of occlusions on optical flow estimation. SKFlow benefits from the super kernels which bring enlarged receptive fields to complement the absent matching information and recover the occluded motions.

Optical flow estimation is a classical yet challenging task in computer vision.

Optical flow estimation is a classical yet challenging task in computer vision. One of the essential factors in accurately predicting optical flow is to alleviate occlusions between frames. However, it is still a thorny problem for current top-performing optical flow estimation methods due to insufficient local evidence to model occluded areas. In this paper, we propose the Super Kernel Flow Network (SKFlow), a CNN architecture to ameliorate the impacts of occlusions on optical flow estimation. SKFlow benefits from the super kernels which bring enlarged receptive fields to complement the absent matching information and recover the occluded motions.

Optical flow estimation is a classical yet challenging task in computer vision. One of the essential factors in accurately predicting optical flow is to alleviate occlusions between frames. However, it is still a thorny problem for current top-performing optical flow estimation methods due to insufficient local evidence to model occluded areas. In this paper, we propose the Super Kernel Flow Network (SKFlow), a CNN architecture to ameliorate the impacts of occlusions on optical flow estimation. SKFlow benefits from the super kernels which bring enlarged receptive fields to complement the absent matching information and recover the occluded motions. We present efficient super kernel designs by utilizing conical connections and hybrid depth-wise convolutions. Extensive experiments demonstrate the effectiveness of SKFlow on multiple benchmarks, especially in the occluded areas. Without pre-trained backbones on ImageNet and with a modest increase in computation, SKFlow achieves compelling performance and ranks $\textbf{1st}$ among currently published methods on the Sintel benchmark. On the challenging Sintel clean and final passes (test), SKFlow surpasses the best-published result in the unmatched areas ($7.96$ and $12.50$) by $9.09\%$ and $7.92\%$. The code is available at \href{https://github.com/littlespray/SKFlow}{https://github.com/littlespray/SKFlow}.

In November, 2015, Google open-sourced its numerical computation library called TensorFlow using data flow graphs. Its flexible implementation and architecture enables you to focus on building the computation graph and deploy the model with little efforts on heterogeous platforms such as mobile devices, hundreds of machines, or thousands of computational devices. TensorFlow is generally very straightforward to use in a sense that most of the researchers in the research area without experience of using this library could understand what's happening behind the code blocks. TensorFlow provides a good backbone for building different shapes of machine learning applications. However, there's a large number of potential users, including some researchers, data scientists, and students who may be familiar with many data science concepts/algorithms already but who never get involved in deep learning research/applications, may found it really hard to start hacking.

In November, 2015, Google open-sourced its numerical computation library called TensorFlow using data flow graphs. Its flexible implementation and architecture enables you to focus on building the computation graph and deploy the model with little efforts on heterogeous platforms such as mobile devices, hundreds of machines, or thousands of computational devices. TensorFlow is generally very straightforward to use in a sense that most of the researchers in the research area without experience of using this library could understand what's happening behind the code blocks. TensorFlow provides a good backbone for building different shapes of machine learning applications. However, there's a large number of potential users, including some researchers, data scientists, and students who may be familiar with many data science concepts/algorithms already but who never get involved in deep learning research/applications, may found it really hard to start hacking.