Advanced Deep Learning-Based Supervised Classification of Multi-Angle Snowflake Camera Images
- 15 June 2021
- journal article
- research article
- Published by American Meteorological Society in Journal of Atmospheric and Oceanic Technology
- Vol. -1 (aop), 1399-1414
- https://doi.org/10.1175/jtech-d-20-0189.1
Abstract
We present improvements over our previous approach to automatic winter hydrometeor classification by means of convolutional neural networks (CNNs), using more data and improved training techniques to achieve higher accuracy on a more complicated dataset than we had previously demonstrated. As an advancement of our previous proof-of-concept study, this work demonstrates broader usefulness of deep CNNs by using a substantially larger and more diverse dataset, which we make publicly available, from many more snow events. We describe the collection, processing, and sorting of this dataset of over 25,000 high-quality multiple-angle snowflake camera (MASC) image chips split nearly evenly between five geometric classes: aggregate, columnar crystal, planar crystal, graupel, and small particle. Raw images were collected over 32 snowfall events between November 2014 and May 2016 near Greeley, Colorado and were processed with an automated cropping and normalization algorithm to yield 224x224 pixel images containing possible hydrometeors. From the bulk set of over 8,400,000 extracted images, a smaller dataset of 14,793 images was sorted by image quality and recognizability (Q&R) using manual inspection. A presorting network trained on the Q&R dataset was applied to all 8,400,000+ images to automatically collect a subset of 283,351 good snowflake images. Roughly 5,000 representative examples were then collected from this subset manually for each of the five geometric classes. With a higher emphasis on in-class variety than our previous work, the final dataset yields trained networks that better capture the imperfect cases and diverse forms that occur within the broad categories studied to achieve an accuracy of 96.2% on a vastly more challenging dataset. We present improvements over our previous approach to automatic winter hydrometeor classification by means of convolutional neural networks (CNNs), using more data and improved training techniques to achieve higher accuracy on a more complicated dataset than we had previously demonstrated. As an advancement of our previous proof-of-concept study, this work demonstrates broader usefulness of deep CNNs by using a substantially larger and more diverse dataset, which we make publicly available, from many more snow events. We describe the collection, processing, and sorting of this dataset of over 25,000 high-quality multiple-angle snowflake camera (MASC) image chips split nearly evenly between five geometric classes: aggregate, columnar crystal, planar crystal, graupel, and small particle. Raw images were collected over 32 snowfall events between November 2014 and May 2016 near Greeley, Colorado and were processed with an automated cropping and normalization algorithm to yield 224x224 pixel images containing possible hydrometeors. From the bulk set of over 8,400,000 extracted images, a smaller dataset of 14,793 images was sorted by image quality and recognizability (Q&R) using manual inspection. A presorting network trained on the Q&R dataset was applied to all 8,400,000+ images to automatically collect a subset of 283,351 good snowflake images. Roughly 5,000 representative examples were then collected from this subset manually for each of the five geometric classes. With a higher emphasis on in-class variety than our previous work, the final dataset yields trained networks that better capture the imperfect cases and diverse forms that occur within the broad categories studied to achieve an accuracy of 96.2% on a vastly more challenging dataset.Keywords
This publication has 15 references indexed in Scilit:
- Physical Dynamics of Ice Crystal GrowthAnnual Review of Materials Research, 2017
- Visual Hull Method for Realistic 3D Particle Shape Reconstruction Based on High-Resolution Photographs of Snowflakes in Free Fall from Multiple ViewsJournal of Atmospheric and Oceanic Technology, 2017
- Dual-Polarized Radar and Surface Observations of a Winter Graupel Shower with Negative Zdr ColumnJournal of Applied Meteorology and Climatology, 2017
- Accurate Characterization of Winter Precipitation Using Multi-Angle Snowflake Camera, Visual Hull, Advanced Scattering Methods and Polarimetric RadarAtmosphere, 2016
- Deep Residual Learning for Image RecognitionPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2016
- ImageNet Large Scale Visual Recognition ChallengeInternational Journal of Computer Vision, 2015
- Hydrometeor classification from two-dimensional video disdrometer dataAtmospheric Chemistry and Physics, 2014
- Ice‐cloud particle habit classification using principal componentsPublished by American Geophysical Union (AGU) ,2012
- Winter Precipitation Microphysics Characterized by Polarimetric Radar and Video Disdrometer Observations in Central OklahomaJournal of Applied Meteorology and Climatology, 2011
- Presenting the Snowflake Video Imager (SVI)Journal of Atmospheric and Oceanic Technology, 2009