DeepSolarEye: Power Loss Prediction and Weakly Supervised Soiling Localization via Fully Convolutional Networks for Solar Panels
Impact of soiling on solar panels is an important and well-studied problem in renewable energy sector. In this paper, we present a novel approach based on fully convolutional networks which takes an RGB image of solar panel and environmental factors (optional) as inputs to predict power loss, soiling localization, and soiling type. In computer vision, predicting localization is a complex task which typically requires human labeled data such as bounding boxes or segmentation masks. Our proposed approach consists of specialized four stages which completely avoids human labeled localization data and only needs panel images with power loss for training. The region of impact area obtained from the localization masks are then classified into soiling types using the webly supervised learning. For superior localization capabilities of convolutional neural networks (CNNs), we introduce a novel bi-directional input-aware fusion (BiDIAF) block that reinforces the input at different levels of CNN to learn input-specific feature maps. Our empirical study shows that BiDIAF improves the power loss prediction accuracy and the localization Jaccard index of ResNet by about 3 further improvement of about 24 supervised manner. Our approach is generalizable and showed promising results on web crawled solar panel images. Additionally, we collected first of it's kind dataset for solar panel image analysis consisting 45,000+ images.
READ FULL TEXT