eduzhai > Applied Sciences > Engineering >

A Retinex based GAN Pipeline to Utilize Paired and Unpaired Datasets for Enhancing Low Light Images

  • Save

... pages left unread,continue reading

Document pages: 6 pages

Abstract: Low light image enhancement is an important challenge for the development ofrobust computer vision algorithms. The machine learning approaches to this havebeen either unsupervised, supervised based on paired dataset or supervisedbased on unpaired dataset. This paper presents a novel deep learning pipelinethat can learn from both paired and unpaired datasets. Convolution NeuralNetworks (CNNs) that are optimized to minimize standard loss, and GenerativeAdversarial Networks (GANs) that are optimized to minimize the adversarial lossare used to achieve different steps of the low light image enhancement process.Cycle consistency loss and a patched discriminator are utilized to furtherimprove the performance. The paper also analyses the functionality and theperformance of different components, hidden layers, and the entire pipeline.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×