DeepSolarEye: Power Loss Prediction and Weakly Supervised Soiling Localization via Fully Convolutional Networks for Solar Panels

Abstract

The impact of soiling on solar panels is an important and well-studied problem in renewable energy sector. In this paper, we present the first convolutional neural network (CNN) based approach for solar panel soiling and defect analysis. Our approach takes an RGB image of solar panel and environmental factors as inputs to predict power loss, soiling localization, and soiling type. In computer vision, localization is a complex task which typically requires manually labeled training data such as bounding boxes or segmentation masks. Our proposed approach consists of specialized four stages which completely avoids localization ground truth and only needs panel images with power loss labels for training. The region of impact area obtained from the predicted localization masks are classified into soiling types using the webly supervised learning. For improving localization capabilities of 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 by about 3% and localization accuracy by about 4%. Our end-to-end model yields further improvement of about 24% on localization when learned in a weakly supervised manner. Our approach is generalizable and showed promising results on web crawled solar panel images. Our system has a frame rate of 22 fps (including all steps) on a NVIDIA TitanX GPU. Additionally, we collected first of it's kind dataset for solar panel image analysis consisting 45,000+ images.

Monitoring photovoltaic soiling: assessment, challenges, and perspectives of current and potential strategies

“…Pictures can be taken from drones (Mehta et al, 2018) or satellites (Supe et al, 2020) and analyzed through standard image processing or machine-learning techniques. In the recent years, researchers have proposed or investigated various image processing methods for soiling detection from modules' pictures (Mehta et al, 2018;Qasem et al, 2017;Supe et al, 2020;Yap et al, 2015). In addition, a novel SIA approach was recently proposed by Yang et al (Yang et al, 2020a.…”

Mask Guided GAN for Density Estimation and Crowd Counting

“…For example, a study [30] uses mask information achieved from the framework to assist RGB and depth images for tracking. Another method [31] adapts the parallel architecture for segmentation and classification with a fully convolutional scheme. Moreover, many applications use deep learning networks (Faster R-CNN or Mask R-CNN) to provide a bounding box or binary mask representing the precise localization of the objects before the later procession, such as pedestrian inspection, obstacle detection, computeraided diagnosis, and vehicle detection [32].…”

A Review of Conventional and Innovative- Sustainable Methods for Cleaning Reflectors in Concentrating Solar Power Plants

“…An investigation performed by [79] used the color and intensity data of digital photographs as indicators to detect soiling. Another method based on convolutional neural networks was proposed to forecast soiling level in [80]. This algorithm uses as inputs RGB images of a soiled surfaces and environmental parameters.…”