SOPNet Method for the Fine-Grained Measurement and Prediction of Precipitation Intensity Using Outdoor Surveillance Cameras

Chih-Wei Lin,Mengxiang Lin,Suhui Yang

doi:10.1109/access.2020.3032430

Chih-Wei Lin, Mengxiang Lin + Show 1 more

Open Access

https://doi.org/10.1109/access.2020.3032430

Copy DOI

Abstract

Surveillance cameras have been widely used in urban environments and are increasingly used in rural ones. Such cameras have mostly been used for security, but they can be applied to the problem of furnishing fine-grained measurements and predictions of precipitation intensity. In this study, we formulated a stacked order-preserving (OP) learning framework to train a network using time-series data. We constructed an OP module, which uses a three-dimensional (3D) convolution operation to extract features with spatial and temporal information and that are associated with ConvLSTM; this feature extraction is used to learn the short-term and OP time-series relationships between features. Furthermore, the OP modules are stacked to form a stacked OP network (SOPNet), which strengthens the relationship between features in long-term time-series image sequences. This SOPNet can be use to obtain fine-grained measurements and predictions of precipitation intensity from images captured by outdoor surveillance cameras. Our main contributions are threefold. First, the SOPNet strengthens the short-term and long-term time-series relationship between features. Second, the SOPNet simultaneously examines spatial and temporal information to measure and predict precipitation intensity. Third, we constructed a precipitation intensity database based on optical images captured by outdoor surveillance cameras. We experimentally evaluated our proposed architecture using our self-collected data set. We found that SOPNet yields better performance and greater accuracy relative to its well-known state-of-the-art counterparts with respect to various metrics.

Highlights

Precipitation refers to the falling of water from the sky to the Earth, and it plays an important role in the hydrological cycle
Where OLSTM is the output of a ConvLSTM cell with 3D feature maps, FLSTM is the operation of a ConvLSTM cell, WLSTM is the convolution kernel of a ConvLSTM cell, bLSTM is the bias of FLSTM, hs−1 is the (s − 1)th hidden gate, and fs−1 is the (s−1)th forget gate
We detail our evaluation of our proposed stacked OP network (SOPNet) against its well-known state-of-the-art counterparts with respect to the measurement of precipitation intensity

Summary

INTRODUCTION

Precipitation refers to the falling of (liquid or solid) water from the sky to the Earth, and it plays an important role in the hydrological cycle. The aforementioned studies have incorporated CNNs into the analysis of video sequences captured by surveillance cameras for object identification and prediction, researchers have yet to formulate an efficient approach that yields fine-grained predictions and measurements of precipitation intensity. Existing methods face problems related to accuracy and spatial and temporal resolution, which hamper their ability to efficiently yield fine-grained predictions and measurements of precipitation intensity. We designed an order-preserving (OP) module to form a stacked OP network (SOPNet) that yields fine-grained forecasts and measurements of precipitation intensity. The designed module, namely the OP module, and the extension operate in conjunction to process short- and long-term time-series information, efficiently improving the accuracy of measurements and forecasts of precipitation intensity.

RELATED WORK

SOPNet

OP MODULE

EXPERIMENTAL RESULTS

CONCLUSION