Weather Classification: A new multi-class dataset, data augmentation approach and comprehensive evaluations of Convolutional Neural Networks

Guerra, J.; Khanam, Zeba; Ehsan, Shoaib; McDonald-Maier, Klaus

doi:10.1109/ahs.2018.8541482

Cited by 29 publications

(8 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another experiment used their own 3 weather images dataset and applied with 10 different CNN models. However, they only achieve a maximum accuracy of 80.70%, which is comparable to our traditional CNN accuracy of 79.8% [37]. There is also another model called DeepCTC, where the model consists of 4 fully connected hidden layers that are fed by 16 neurons from GOES-16 ABI as input and 9 SoftMax nodes as clouds categories.…”

Section: Discussionmentioning

confidence: 80%

Improving Accuracy of Cloud Images using DenseNet-VGG19

Fitriana¹,

Arifa²,

Prasetiadi³

et al. 2023

Int. J. Adv. Sci. Eng. Inf. Technol.

View full text Add to dashboard Cite

Weather classification has become a significant challenge due to the unpredictable nature of climate conditions. For farmers, predicting the start of the rainy season is very important. This is because it is related to the cost factor that must be incurred, and also, the waiting time for the harvest has an effect if the weather is not supportive. Farmers also have to prepare seeds for the start of their farming. Therefore, farmers who start nurseries early in the rainy season will miss significant planting time. Based on these problems, this study uses a convolutional neural network (CNN) for weather classification using cloud imagery. CNN is shown to classify different spectro-temporal features of sound and is thus suitable for cloud image classification. We collect cloud image data using secondary data. Our model used a layer based on the convolution CNN architecture with a pooling layer and a solid layer as the output layer. The cloud dataset used is 1230 data divided into five classes, namely cloudy, foggy, rain, shine, and sunrise, which we use to train our model in research for the feature extraction process using DenseNet and VGG19. We use two types of classification, namely fully connected and Global Average Pooling (GAP). Our model can achieve a classification accuracy of 90.8% DenseNet-Fully Connected from our training process. From our testing process, our model can reach 95.7% using DenseNet-Fully Connected classification accuracy. Thus, the CNN model proved very accurate in classifying cloud images.

show abstract

Section: Discussionmentioning

confidence: 80%

Improving Accuracy of Cloud Images using DenseNet-VGG19

Fitriana¹,

Arifa²,

Prasetiadi³

et al. 2023

Int. J. Adv. Sci. Eng. Inf. Technol.

View full text Add to dashboard Cite

show abstract

“…The real-world images are mainly from rain fog snow (RFS), 71 multi-class weather dataset (MWD), 72 reside-β, 73 snow removal in realistic scenario (SRRS), 9 Image2Weather, 10 dataset2 74 and real_world_rain_dataset 75 (RWRD). Given that the purpose of this paper is to remove the degradation caused by the bad weather conditions, our method’s judgment principle for the image weather type relies on whether the occurring weather has caused the corresponding degradation.…”

Section: Methodsmentioning

confidence: 99%

Framework of degraded image restoration and simultaneous localization and mapping for multiple bad weather conditions

Dong

Guo

et al. 2023

Opt. Eng.

View full text Add to dashboard Cite

Image degradation caused by bad weather, such as haze, rain, and snow, significantly degrades a vision system's performance, imposing challenges to the autonomous motion of mobile robots. Therefore, it is crucial to restore the degraded images for mobile robots operating outdoors. Spurred by this concern, a framework that restores degraded images and affords simultaneous localization and mapping (SLAM) under multiple bad weather conditions is presented. Specifically, the developed architecture combines the advantages of convolutional neural network features and weather features. It improves the accuracy of identifying the degradation type by developing a weather inference module based on ensemble learning. According to the internal mechanism of image degradation, a degraded image restoration method is proposed utilizing physical models, with a subsequent operation refining the preliminary restored results. To improve our method's adaptability to real scenes, unpaired real-world weather images are introduced into the degradation removal algorithm through generative adversarial networks. The proposed weather persistent assumption combines weather inference and degradation restoration modules in the SLAM system to capture multiple weather conditions, which improves the system's accuracy and running speed. Comprehensive experiments evaluating the developed framework and its components highlight that the proposed weather inference and degraded image restoration methods achieve a highly appealing effect. The final experimental results demonstrate that our framework autonomously identifies weather types, triggers the corresponding restoration, and realizes accurate localization in multiple bad weather conditions. It affords SLAM accuracy under multiple bad weather conditions that is close to raw SLAM in clear weather.

show abstract

“…In order to increase the number of training images and cover unexpected inputs, data augmentation can be used on the input dataset to increase the number of inputs. Villarreal Guerra et al [15] used multiple superpixel groups on weather datasets with multiple CNN models to compare their performance. The superpixels on the sunny dataset improved the performance of every model.…”

Section: Related Workmentioning

confidence: 99%

Hierarchical Structure for Semantic Segmentation of Roadway Images with Limited Visibility Using Deep Artificial Neural Networks

An,

Mohan,

Leppo

et al. 2023

Preprint

View full text Add to dashboard Cite

The field of autonomous driving leaves minimal margins for error. Ensuring that self-driving vehicles possess the ability to accurately perceive their surroundings, even amidst conditions of limited visibility, is of utmost importance. We propose a novel approach to enhance the precision of object detection on the road during limited visibility driving or roadway conditions. The initial step involves the classification of the driving condition of an input image, and then, the corresponding semantic segmentation model will process the image to distinguish objects. Our dataset consists of roadway images depicting 20 distinct objects amidst adverse limited visibility conditions. The experimental results validate our approach, with the proposed method displaying quality accuracy levels for training, validation, and testing data. Our classification model achieved 100% accuracy. Particularly, the proposed methods achieved final mean IoU scores of 57.3%, 32.0%, 49.4%, and 47.8%, respectively, for FOG, NIGHT, RAIN, and SNOW conditions when using the U-NET model for segmentation. These mean IoU results are better than the traditional nonhierarchical training methods, which utilize the same U-NET structure.

show abstract

Weather Classification: A new multi-class dataset, data augmentation approach and comprehensive evaluations of Convolutional Neural Networks

Cited by 29 publications

References 20 publications

Improving Accuracy of Cloud Images using DenseNet-VGG19

Improving Accuracy of Cloud Images using DenseNet-VGG19

Framework of degraded image restoration and simultaneous localization and mapping for multiple bad weather conditions

Hierarchical Structure for Semantic Segmentation of Roadway Images with Limited Visibility Using Deep Artificial Neural Networks

Contact Info

Product

Resources

About