“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

While it is nearly effortless for humans to quickly assess the perceptual similarity between two images, the underlying processes are thought to be quite complex. Despite this, the most widely used perceptual metrics today, such as PSNR and SSIM, are simple, shallow functions, and fail to account for many nuances of human perception. Recently, the deep learning community has found that features of the VGG network trained on ImageNet classification has been remarkably useful as a training loss for image synthesis. But how perceptual are these so-called "perceptual losses"? What elements are critical for their success? To answer these questions, we introduce a new dataset of human perceptual similarity judgments. We systematically evaluate deep features across different architectures and tasks and compare them with classic metrics. We find that deep features outperform all previous metrics by large margins on our dataset. More surprisingly, this result is not restricted to ImageNet-trained VGG features, but holds across different deep architectures and levels of supervision (supervised, self-supervised, or even unsupervised). Our results suggest that perceptual similarity is an emergent property shared across deep visual representations.

The Unreasonable Effectiveness of Deep Features as a Perceptual Metric

/pdf/the-unreasonable-effectiveness-of-deep-features-as-a-10k66u5qas.pdf

This paper reviews the first challenge on single image super-resolution (restoration of rich details in an low resolution image) with focus on proposed solutions and results. A new DIVerse 2K resolution image dataset (DIV2K) was employed. The challenge had 6 competitions divided into 2 tracks with 3 magnification factors each. Track 1 employed the standard bicubic downscaling setup, while Track 2 had unknown downscaling operators (blur kernel and decimation) but learnable through low and high res train images. Each competition had ∽100 registered participants and 20 teams competed in the final testing phase. They gauge the state-of-the-art in single image super-resolution.

/pdf/ntire-2017-challenge-on-single-image-super-resolution-2lwnw6xe8t.pdf

NTIRE 2017 Challenge on Single Image Super-Resolution: Methods and Results

In single image deblurring, the "coarse-to-fine" scheme, i.e. gradually restoring the sharp image on different resolutions in a pyramid, is very successful in both traditional optimization-based methods and recent neural-network-based approaches. In this paper, we investigate this strategy and propose a Scale-recurrent Network (SRN-DeblurNet) for this deblurring task. Compared with the many recent learning-based approaches in [25], it has a simpler network structure, a smaller number of parameters and is easier to train. We evaluate our method on large-scale deblurring datasets with complex motion. Results show that our method can produce better quality results than state-of-the-arts, both quantitatively and qualitatively.

/pdf/scale-recurrent-network-for-deep-image-deblurring-45johrtaik.pdf

Scale-Recurrent Network for Deep Image Deblurring

Motion blur from camera shake is a major problem in videos captured by hand-held devices. Unlike single-image deblurring, video-based approaches can take advantage of the abundant information that exists across neighboring frames. As a result the best performing methods rely on the alignment of nearby frames. However, aligning images is a computationally expensive and fragile procedure, and methods that aggregate information must therefore be able to identify which regions have been accurately aligned and which have not, a task that requires high level scene understanding. In this work, we introduce a deep learning solution to video deblurring, where a CNN is trained end-to-end to learn how to accumulate information across frames. To train this network, we collected a dataset of real videos recorded with a high frame rate camera, which we use to generate synthetic motion blur for supervision. We show that the features learned from this dataset extend to deblurring motion blur that arises due to camera shake in a wide range of videos, and compare the quality of results to a number of other baselines.

/pdf/deep-video-deblurring-for-hand-held-cameras-3u7kprauh3.pdf

Deep Video Deblurring for Hand-Held Cameras

We present an end-to-end image processing framework for time-of-flight (ToF) cameras. Existing ToF image processing pipelines consist of a sequence of operations including modulated exposures, denoising, phase unwrapping and multipath interference correction. While this cascaded modular design offers several benefits, such as closed-form solutions and power-efficient processing, it also suffers from error accumulation and information loss as each module can only observe the output from its direct predecessor, resulting in erroneous depth estimates. We depart from a conventional pipeline model and propose a deep convolutional neural network architecture that recovers scene depth directly from dual-frequency, raw ToF correlation measurements. To train this network, we simulate ToF images for a variety of scenes using a time-resolved renderer, devise depth-specific losses, and apply normalization and augmentation strategies to generalize this model to real captures. We demonstrate that the proposed network can efficiently exploit the spatio-temporal structures of ToF frequency measurements, and validate the performance of the joint multipath removal, denoising and phase unwrapping method on a wide range of challenging scenes.

/pdf/deep-end-to-end-time-of-flight-imaging-3v2t0g6n1z.pdf

Deep End-to-End Time-of-Flight Imaging

Diffractive optical elements (DOE) show great promise for imaging optics that are thinner and more lightweight than conventional refractive lenses while preserving their light efficiency. Unfortunately, severe spectral dispersion currently limits the use of DOEs in consumer-level lens design. In this article, we jointly design lightweight diffractive-refractive optics and post-processing algorithms to enable imaging under white light illumination. Using the Fresnel lens as a general platform, we show three phase-plate designs, including a super-thin stacked plate design, a diffractive-refractive-hybrid lens, and a phase coded-aperture lens. Combined with cross-channel deconvolution algorithm, both spherical and chromatic aberrations are corrected. Experimental results indicate that using our computational imaging approach, diffractive-refractive optics is an alternative candidate to build light efficient and thin optics for white light imaging.

Computational imaging using lightweight diffractive-refractive optics.

Although motion blur and rolling shutter deformations are closely coupled artifacts in images taken with CMOS image sensors, the two phenomena have so far mostly been treated separately, with deblurring algorithms being unable to handle rolling shutter wobble, and rolling shutter algorithms being incapable of dealing with motion blur. We propose an approach that delivers sharp and undistorted output given a single rolling shutter motion blurred image. The key to achieving this is a global modeling of the camera motion trajectory, which enables each scanline of the image to be deblurred with the corresponding motion segment. We show the results of the proposed framework through experiments on synthetic and real data.

/pdf/rolling-shutter-motion-deblurring-24g3ikzd6x.pdf

Rolling shutter motion deblurring

Motion blur from camera shake is a major problem in videos captured by hand-held devices. Unlike single-image deblurring, video-based approaches can take advantage of the abundant information that exists across neighboring frames. As a result the best performing methods rely on aligning nearby frames. However, aligning images is a computationally expensive and fragile procedure, and methods that aggregate information must therefore be able to identify which regions have been accurately aligned and which have not, a task which requires high level scene understanding. In this work, we introduce a deep learning solution to video deblurring, where a CNN is trained end-to-end to learn how to accumulate information across frames. To train this network, we collected a dataset of real videos recorded with a high framerate camera, which we use to generate synthetic motion blur for supervision. We show that the features learned from this dataset extend to deblurring motion blur that arises due to camera shake in a wide range of videos, and compare the quality of results to a number of other baselines.

Shuochen Su

Papers

Deep Video Deblurring for Hand-Held Cameras

Deep End-to-End Time-of-Flight Imaging

Computational imaging using lightweight diffractive-refractive optics.

Rolling shutter motion deblurring

Deep Video Deblurring.