See clearer at night: towards robust nighttime semantic segmentation through day-night image conversion

Abstract

In recent years, intelligent driving navigation and security monitoring have made considerable progress with the help of deep Convolutional Neural Networks (CNNs). As one of the state-of-the-art perception approaches, semantic segmentation unifies distinct detection tasks widely desired by both autonomous driving and security monitoring. Currently, semantic segmentation shows remarkable efficiency and reliability in standard scenarios such as daytime scenes with favorable illumination conditions. However, in face of adverse conditions such as the nighttime, semantic segmentation loses its accuracy significantly. One of the main causes of the problem is the lack of sufficient annotated segmentation datasets of nighttime scenes. In this paper, we propose a framework to alleviate the accuracy decline when semantic segmentation is taken to adverse conditions by using Generative Adversarial Networks (GANs). To bridge the daytime and nighttime image domains, we made key observation that compared to datasets in adverse conditions, there are considerable amount of segmentation datasets in standard conditions such as BDD and our collected ZJU datasets. Our GAN-based nighttime semantic segmentation framework includes two methods. In the first method, GANs were used to translate nighttime images to the daytime, thus semantic segmentation can be performed using robust models already trained on daytime datasets. In another method, we use GANs to translate different ratio of daytime images in the dataset to the nighttime but still with their labels. In this sense, synthetic nighttime segmentation datasets can be generated to yield models prepared to operate at nighttime conditions robustly. In our experiment, the later method significantly boosts the performance at the nighttime evidenced by quantitative results using Intersection over Union (IoU) and Pixel Accuracy (Acc). We show that the performance varies with respect to the proportion of synthetic nighttime images in the dataset, where the sweet spot corresponds to most robust performance across the day and night. The proposed framework not only makes contribution to the optimization of visual perception in intelligent vehicles, but also can be applied to diverse navigational assistance systems.

Deep multimodal fusion for semantic image segmentation: A survey

“…Zhao et al [54] presented an image cascade network, known as ICNet, that incorporates multi-resolution branches under proper label guidance. In more recent years, semantic segmentation for adverse weather conditions [55,56] and nighttime [57,58] has also been addressed to perform the generalization capacity and robustness of deep learning models. Figure 2 shows the number of papers about segmentation published in the past decade.…”

Modified Perceptual Cycle Generative Adversarial Network-Based Image Enhancement for Improving Accuracy of Low Light Image Segmentation

“…To overcome these drawbacks, various semantic segmentation methods have been studied in extremely low light or nighttime environments [1][2][3][4][5][6][7][8][9][10][11][12][13][14].…”

“…Due to these problems, semantic segmentation is extremely difficult in low light environment and improving the performance is a challenging problem. Accordingly, various methods for object segmentation have been studied in low light or nighttime environments [1][2][3][4][5][6][7][8][9][10][11][12][13][14]. In handcrafted feature-based segmentation studies [1][2][3][4][5][6][7][8][9], segmentation is performed by classifying the single object region as foreground and the rest of the areas as background in a low light environment.…”

“…In deep feature-based segmentation studies [10][11][12][13][14], the methods for performing segmentation of multiple objects in a low light environment have been researched. For nonenhancement based methods [10][11][12][13], low light or nighttime images and a manually annotated label are used for training a CNN model. The features that can classify each object accurately are automatically extracted and trained to perform segmentation.…”

Semantic Segmentation With Low Light Images by Modified CycleGAN-Based Image Enhancement

“…However, a dataset with various illumination changes was required, the training time was long, and the training complexity was high. Sun et al [14] performed night-to-day image conversion using CycleGAN, but the generated image quality was poor and the authors did not show quantitative experimental results. The second proposed method, CycleGANbased nighttime image augmentation method, increases the number of training data by generating synthetic nighttime images, and improves segmentation accuracy by training the proposed segmentation network.…”