Enhanced CycleGAN Dehazing Network

Single image dehazing is a challenging problem, and it is far from solved. Most current solutions require paired image datasets that include both hazy images and their corresponding haze-free ground-truth. How- ever, in reality lighting conditions and other factors can produce a range of haze-free images that can serve as ground truth for a hazy image, and a single ground truth image cannot capture that range. This limits the scalability and practicality of paired methods in real-world applications. In this paper, we focus on unpaired single image dehazing and reduce the image dehazing problem to an unpaired image-to-image translation and propose an Enhanced CycleGAN Dehazing Network (ECDN). We enhance CycleGAN from different angles for the dehazing purpose. We employ a global-local discriminator structure to deal with spatially varying haze. We define self-regularized color loss and utilize it along with perceptual loss to generate more realistic and visually pleasing images. We use an encoder-decoder archi- tecture with residual blocks in the generator with skip connections so that the network better preserves the details. Through an ablation study, we demonstrate the effectiveness of different modules in the performance of the proposed network. Our extensive experiments over two benchmark datasets show that our network outperforms previous work in terms of PSNR and SSIM.

[Paper] [Code]

Evaluating Single Image Dehazing Methods Under Realistic Sunlight Haze.

Haze can degrade the visibility and the image quality dras- tically, thus degrading the performance of computer vision tasks such as object detection. Single image dehazing is a challenging and ill-posed problem, despite being widely studied. Most existing methods assume that haze has a uniform/homogeneous distribution and haze can have a single color, i.e. grayish white color similar to smoke, while in reality haze can be distributed non-uniformly with different patterns and col- ors. In this paper, we focus on haze created by sunlight as it is one of the most prevalent types of haze in the wild. Sunlight can generate non- uniformly distributed haze with drastic density changes due to sun rays and also a spectrum of haze color due to sunlight color changes during the day. This presents a new challenge to image dehazing methods. For these methods to be practical, this problem needs to be addressed. To quantify the challenges and assess the performance of these methods, we present a sunlight haze benchmark dataset, Sun-Haze, containing 107 hazy images with different types of haze created by sunlight having a variety of intensity and color. We evaluate a representative set of state- of-the-art image dehazing methods on this benchmark dataset in terms of standard metrics such as PSNR, SSIM, CIEDE2000, PI and NIQE. Our results provide information on limitations of current methods, and on the practicality of their underlying assumptions.

[Paper]

A Pipeline for Automated Face Dataset Creation from Unlabeled Images.

Computer vision tasks are very dataset dependent and with the emerging growth of image processing tasks and their applications, there is a huge demand for more datasets in terms of variety or size. Considering many face datasets are assembled largely manually, the process of manual dataset construction could be very time- consuming and labor intensive. In addition, some face datasets are constructed automatically over the past few years, but they are all constructed with previous knowledge of labels to some extent. In this work, we focus on automated face dataset generation from unlabeled images. We present a novel and effective pipeline for gen- erating face datasets automatically from unlabeled and raw data for face-related tasks. We evaluated our pipeline on several datasets and it achieved a significant improvement compared to clustering-only approaches, which shows its potential towards practical solutions for automated face dataset generation.

[Paper]