ACLNet: An Attention and Clustering-based Cloud Segmentation Network

TL;DR

ACLNet is a novel deep learning model that combines neural networks and clustering techniques to improve cloud segmentation accuracy in ground images across different lighting conditions.

Contribution

It introduces a new architecture integrating EfficientNet, ASPP, GAM, and k-means clustering for enhanced cloud boundary detection.

Findings

Lower error rate than existing models

Higher recall and F1-score in cloud segmentation

Effective for both daytime and nighttime images

Abstract

We propose a novel deep learning model named ACLNet, for cloud segmentation from ground images. ACLNet uses both deep neural network and machine learning (ML) algorithm to extract complementary features. Specifically, it uses EfficientNet-B0 as the backbone, "`a trous spatial pyramid pooling" (ASPP) to learn at multiple receptive fields, and "global attention module" (GAM) to extract finegrained details from the image. ACLNet also uses k-means clustering to extract cloud boundaries more precisely. ACLNet is effective for both daytime and nighttime images. It provides lower error rate, higher recall and higher F1-score than state-of-art cloud segmentation models. The source-code of ACLNet is available here: https://github.com/ckmvigil/ACLNet.