Abstract: Images obtained in real-world low-light conditions are not only low in brightness, but they also suffer from many other types of degradation, such as color bias, unknown noise, detail loss and halo artifacts. In this paper, we propose a very fast deep learning framework called Bringing the Lightness (denoted as BLNet) that consists of two U-Nets with a series of well-designed loss functions to tackle all of the above degradations. Based on Retinex Theory, the decomposition net in our model can decompose low-light images into reflectance and illumination and remove noise in the reflectance during the decomposition phase. We propose a Noise and Color Bias Control module (NCBC Module) that contains a convolutional neural network and two loss functions (noise loss and color loss). This module is only used to calculate the loss functions during the training phase, so our method is very fast during the test phase. This module can smooth the reflectance to achieve the purpose of noise removal while preserving details and edge information and controlling color bias. We propose a network that can be trained to learn the mapping between low-light and normal-light illumination and enhance the brightness of images taken in low-light illumination. We train and evaluate the performance of our proposed model over the real-world Low-Light (LOL) dataset), and we also test our model over several other frequently used datasets (LIME, DICM and MEF datasets). We conduct extensive experiments to demonstrate that our approach achieves a promising effect with good rubustness and generalization and outperforms many other state-of-the-art methods qualitatively and quantitatively. Our method achieves high speed because we use loss functions instead of introducing additional denoisers for noise removal and color correction. The code and model are available at https://github.com/weixinxu666/BLNet.

Rocket Launching: A Universal and Efficient Framework for Training Well-Performing Light Net

Click-Through Rate Prediction Algorithm Based on Modeling of Implicit High-Order Feature Importance

BOOST: Block Minifloat-Based On-Device CNN Training Accelerator with Transfer Learning

NetBooster: Empowering Tiny Deep Learning By Standing on the Shoulders of Deep Giants

Rocket Image Classification Based on Deep Convolutional Neural Network

BLNet: A Fast Deep Learning Framework for Low-Light Image Enhancement with Noise Removal and Color Restoration

Optimal Reusable Rocket Landing Guidance: A Cutting-Edge Approach Integrating Scientific Machine Learning and Enhanced Neural Networks

Visualizing and Understanding Deep Neural Networks in CTR Prediction

CowClip: Reducing CTR Prediction Model Training Time from 12 hours to 10 minutes on 1 GPU

High-Performance Large-Scale Image Recognition Without Normalization

Gradient Boosting Neural Networks: GrowNet

Lightweight Network Research Based on Deep Learning

A multi-head attention network with adaptive meta-transfer learning for RUL prediction of rocket engines

Energy-efficient and Robust Cumulative Training with Net2Net Transformation

Iterative Boosting Deep Neural Networks for Predicting Click-Through Rate

Understanding and Accelerating Neural Architecture Search with Training-Free and Theory-Grounded Metrics

Communication-Efficient TeraByte-Scale Model Training Framework for Online Advertising

AccuLiteFastNet: A Remote Sensing Object Detection Model Combining High Accuracy, Lightweight Design, and Fast Inference Speed

Boosting Tail Neural Network for Realtime Custom Keyword Spotting

Glance and Focus: a Dynamic Approach to Reducing Spatial Redundancy in Image Classification

MultiRocket: multiple pooling operators and transformations for fast and effective time series classification