Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

HipyrNet: Hypernet-Guided Feature Pyramid network for mixed-exposure correction

Shaurya Singh Rathore, Aravind Shenoy, Krish Didwania, Aditya Kasliwal, Ujjwal Verma

TL;DR

HipyrNet addresses mixed-exposure correction by embedding a HyperNetwork within a Laplacian Pyramid framework to dynamically generate the decomposition kernel for each input. The architecture blends multiscale feature translation with input-conditioned kernels, enabling adaptive restoration of both under- and overexposed regions. A composite loss combines pixel, adversarial, and kernel-prediction terms to improve fidelity while maintaining adaptability across diverse exposures. Evaluations on the SICE, SICE-Grad, and SICE-Mix datasets show state-of-the-art PSNR/SSIM performance, with notable improvements over prior methods, highlighting the practical value of dynamic, per-image kernel generation for exposure correction and paving the way for real-time, adaptive image translation tasks.

Abstract

Recent advancements in image translation for enhancing mixed-exposure images have demonstrated the transformative potential of deep learning algorithms. However, addressing extreme exposure variations in images remains a significant challenge due to the inherent complexity and contrast inconsistencies across regions. Current methods often struggle to adapt effectively to these variations, resulting in suboptimal performance. In this work, we propose HipyrNet, a novel approach that integrates a HyperNetwork within a Laplacian Pyramid-based framework to tackle the challenges of mixed-exposure image enhancement. The inclusion of a HyperNetwork allows the model to adapt to these exposure variations. HyperNetworks dynamically generates weights for another network, allowing dynamic changes during deployment. In our model, the HyperNetwork employed is used to predict optimal kernels for Feature Pyramid decomposition, which enables a tailored and adaptive decomposition process for each input image. Our enhanced translational network incorporates multiscale decomposition and reconstruction, leveraging dynamic kernel prediction to capture and manipulate features across varying scales. Extensive experiments demonstrate that HipyrNet outperforms existing methods, particularly in scenarios with extreme exposure variations, achieving superior results in both qualitative and quantitative evaluations. Our approach sets a new benchmark for mixed-exposure image enhancement, paving the way for future research in adaptive image translation.

HipyrNet: Hypernet-Guided Feature Pyramid network for mixed-exposure correction

TL;DR

HipyrNet addresses mixed-exposure correction by embedding a HyperNetwork within a Laplacian Pyramid framework to dynamically generate the decomposition kernel for each input. The architecture blends multiscale feature translation with input-conditioned kernels, enabling adaptive restoration of both under- and overexposed regions. A composite loss combines pixel, adversarial, and kernel-prediction terms to improve fidelity while maintaining adaptability across diverse exposures. Evaluations on the SICE, SICE-Grad, and SICE-Mix datasets show state-of-the-art PSNR/SSIM performance, with notable improvements over prior methods, highlighting the practical value of dynamic, per-image kernel generation for exposure correction and paving the way for real-time, adaptive image translation tasks.

Abstract

Recent advancements in image translation for enhancing mixed-exposure images have demonstrated the transformative potential of deep learning algorithms. However, addressing extreme exposure variations in images remains a significant challenge due to the inherent complexity and contrast inconsistencies across regions. Current methods often struggle to adapt effectively to these variations, resulting in suboptimal performance. In this work, we propose HipyrNet, a novel approach that integrates a HyperNetwork within a Laplacian Pyramid-based framework to tackle the challenges of mixed-exposure image enhancement. The inclusion of a HyperNetwork allows the model to adapt to these exposure variations. HyperNetworks dynamically generates weights for another network, allowing dynamic changes during deployment. In our model, the HyperNetwork employed is used to predict optimal kernels for Feature Pyramid decomposition, which enables a tailored and adaptive decomposition process for each input image. Our enhanced translational network incorporates multiscale decomposition and reconstruction, leveraging dynamic kernel prediction to capture and manipulate features across varying scales. Extensive experiments demonstrate that HipyrNet outperforms existing methods, particularly in scenarios with extreme exposure variations, achieving superior results in both qualitative and quantitative evaluations. Our approach sets a new benchmark for mixed-exposure image enhancement, paving the way for future research in adaptive image translation.
Paper Structure (15 sections, 6 equations, 3 figures, 4 tables)

This paper contains 15 sections, 6 equations, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Image contrast Enhancement
  4. Feature Pyramid Networks
  5. Laplacian Pyramid Translation Network
  6. HyperNetworks
  7. Architecture
  8. Feature Pyramid
  9. Loss Function
  10. HyperNet Architecture
  11. Experiments
  12. Datasets
  13. Hyperparameter Tuning
  14. Gan Loss Functions
  15. Results

Figures (3)

  • Figure 1: A schematic of the proposed model. The figure represents the Hypernet, and the following model uses the predicted kernel to translate the underexposed image. UTM (Upper Translational Module) and LTM (Lower Translational Module) translate the feature pyramid. The exact architecture of the Hypernet is denoted in Fig. \ref{['fig:hypernet']}. The image is taken from the SICE Dataset 8259342.
  • Figure 2: Expanded Schematic of the Hypernet. The figure represents the Hypernet architecture in detail and shows each of the convolution blocks and their corresponding output channels
  • Figure 3: Visualizations of images from the SICE 8259342 and the SICE Grad and SICE Mix zheng2024lowlightimagevideoenhancement Dataset. The first row refers to an overexposed image from SICE, the second row is of an underexposed image from SICE, and the last two rows refer to SICE Grad and SICE Mi,x respectively