Efficient Shallow Network for River Ice Segmentation

Abstract

River ice segmentation, used for surface ice concentration estimation, is important for validating river processes and ice-formation models, predicting ice jam and flooding risks, and managing water supply and hydroelectric power generation. Furthermore, discriminating between anchor ice and frazil ice is an important factor in understanding sediment transport and release events. Modern deep learning techniques have proved to deliver promising results; however, they can show poor generalization ability and can be inefficient when hardware and computing power is limited. As river ice images are often collected in remote locations by unmanned aerial vehicles with limited computation power, we explore the performance-latency trade-offs for river ice segmentation. We propose a novel convolution block inspired by both depthwise separable convolutions and local binary convolutions giving additional efficiency and parameter savings. Our novel convolution block is used in a shallow architecture which has 99.9% fewer trainable parameters, 99% fewer multiply–add operations, and 69.8% less memory usage than a UNet, while achieving virtually the same segmentation performance. We find that the this network trains fast and is able to achieve high segmentation performance early in training due to an emphasis on both pixel intensity and texture. When compared to very efficient segmentation networks such as LR-ASPP with a MobileNetV3 backbone, we achieve good performance (mIoU of 64) 91% faster during training on a CPU and an overall mIoU that is 7.7% higher. We also find that our network is able to generalize better to new domains such as snowy environments.

Waved-Shape Accumulation of Ice Jam—Analysis and Experimental Study

“…During the ice-pans floating period, the surface layer of flowing water lost too much heat due to the decrease in air temperature [19]. On the other side, the turbulent mixing process of flowing water causes the heat exchange between the water surface layer and the main water body.…”

Assessing Ice Break-Up Trends in Slave River Delta through Satellite Observations and Random Forest Modeling

“…A study by Sobiech and Dierking [39] evaluated the performance of kmeans classification on lakes and river channels of the central Lena Delta and showed that it is comparable to that of a fixed-threshold approach. Neural networks, support vector machines (SVMs), and RF classifiers are popular examples of supervised learning methods used in river and lake ice classification [40][41][42]. A review study by Belgiu and Drăguţ [43] reports that RF classifiers outperform artificial neural network classifiers in terms of classification accuracy and provide slightly better results than SVMs for highdimensional input data such as hyperspectral imagery [43,44].…”