The proposed neural network is motivated by the low-rank property of pseudo-inverse kernels.

Computer vision tasks often have side information available that is helpful to solve the task. For example, for crowd counting, the camera perspective (e.g., camera angle and height) gives a clue about the appearance and scale of people in the scene. While side information has been shown to be useful for counting systems using traditional hand-crafted features, it has not been fully utilized in counting systems based on deep learning.

Neural Information Processing Systems http://nips.cc/

First provide a summary of the paper, and then address the following criteria: Quality, clarity, originality and significance. The paper proposes to use a deep convolutional neural network for denoising images by generating a lot of noisy and noiseless image pairs using a synthetic blurring process. The proposed method achieves good results on a number of image deblurring tasks. The idea is simple and elegant. It is observed that a 2D deconvolution, which is an inverse of the convolution operator, is itself a 2D convolution operator, albeit with a very large support.

Neural Information Processing Systems http://nips.cc/

Many fundamental image-related problems involve deconvolution operators. Real blur degradation seldom complies with an deal linear convolution model due to camera noise, saturation, image compression, to name a few. Instead of perfectly modeling outliers, which is rather challenging from a generative model perspective, we develop a deep convolutional neural network to capture the characteristics of degradation. We note directly applying existing deep neural networks does not produce reasonable results. Our solution is to establish the connection between traditional optimization-based schemes and a neural network architecture where a novel, separable structure is introduced as a reliable support for robust deconvolution against artifacts. Our network contains two submodules, both trained in a supervised manner with proper initialization. They yield decent performance on non-blind image deconvolution compared to previous generative-model based methods.

Many fundamental image-related problems involve deconvolution operators. Real blur degradation seldom complies with an ideal linear convolution model due to camera noise, saturation, image compression, to name a few. Instead of perfectly modeling outliers, which is rather challenging from a generative model perspective, we develop a deep convolutional neural network to capture the characteristics of degradation. We note directly applying existing deep neural networks does not produce reasonable results. Our solution is to establish the connection between traditional optimization-based schemes and a neural network architecture where a novel, separable structure is introduced as a reliable support for robust deconvolution against artifacts. Our network contains two submodules, both trained in a supervised manner with proper initialization. They yield decent performance on non-blind image deconvolution compared to previous generative-model based methods.

Many fundamental image-related problems involve deconvolution operators. Real blur degradation seldom complies with an deal linear convolution model due to camera noise, saturation, image compression, to name a few. Instead of perfectly modeling outliers, which is rather challenging from a generative model perspective, we develop a deep convolutional neural network to capture the characteristics of degradation. We note directly applying existing deep neural networks does not produce reasonable results. Our solution is to establish the connection between traditional optimization-based schemes and a neural network architecture where a novel, separable structure is introduced as a reliable support for robust deconvolution against artifacts.

Summary of the Paper: This work proposes to use adaptive convolutions (also called'cross convolutions') to incorporate side information (e.g., camera angle) into CNN architectures for vision tasks (e.g., crowd counting). The filter weights in each adaptive convolution layer are predicted using a separate neural network (one network for each set of filter weights) with is a multi-layer perceptron. This network is referred to as'Filter Manifold Network' which takes the auxiliary side information as input and predicts the filter weights. Experiments on three vision tasks of crowd counting, digit recognition and image deconvolution indicate the potential of the proposed technique for incorporating auxiliary information. In addition, this paper contributes a new dataset for crowd counting with different camera heights and angles.

This paper extends the non-blind deconvolution in [1]. The author proposes to use a generalized low-rank approximation method to model a wide range of blur kernels and train the non-blind deconvolution network with the low-rank representation. Strength: - Unlike the model in [1] that requires a specific training (or fine tuning) for each kernel, the proposed model can handle different blur kernels. Weakness: - The most important motivation of this paper is to make the learning-based CNN model being able to handle different kernels. I did not find the reason for using the low-rank approximation-based method.