Application of Deep Convolutional Neural Networks in Detecting

  • Slides: 14
Download presentation
Application of Deep Convolutional Neural Networks in Detecting Severe Weather Events from Radar Reflectivity

Application of Deep Convolutional Neural Networks in Detecting Severe Weather Events from Radar Reflectivity Images Rachel Under the supervision of Professor Fuqing Zhang and Professor Jia Li May 10, 2019

High-level Visual Signals in Weather Images Hook Echo, Velocity Couplet, The V-Notch, or “Flying

High-level Visual Signals in Weather Images Hook Echo, Velocity Couplet, The V-Notch, or “Flying Eagle”, The Debris Ball, etc. images credit to: https: //www. ustornadoes. com/2013/02/14/understanding-basic-tornadic-radar-signatures/

Motivation 1. Some of the human-designed visual features are hard to interpret. 2. Severe

Motivation 1. Some of the human-designed visual features are hard to interpret. 2. Severe weather detection relies much on experience. 3. Deep learning has the ability to explore the underneath pattern of the images. Solution: Train a Deep Neural Network to detect severe weather events.

Regional Convolutional Neural Network (RCNN)

Regional Convolutional Neural Network (RCNN)

Part 1: Convolutional Neural Network (CNN) Function: output the visual features of the images.

Part 1: Convolutional Neural Network (CNN) Function: output the visual features of the images. Method: Fix the first 10/24/30 layers of CNN and retrain with radar reflectivity images. Output: feature map

Part 2: Region Proposal Network (RPN) Function: select regions from the region candidates generate

Part 2: Region Proposal Network (RPN) Function: select regions from the region candidates generate from the anchors and image features. Adjust anchor positions. Output: about 2000 positive+negative regions.

Part 3: Region of Interest (ROI) Pooling Function: further select regions of interest on

Part 3: Region of Interest (ROI) Pooling Function: further select regions of interest on region candidates. Adjust positions of the regions. Output: 128 positive+negative regions. Loss: sum of the cross entropy in RPN and ROI pooling stage.

Dataset 1. The historical radar reflection images from 2008 to 2017. 2. The severe

Dataset 1. The historical radar reflection images from 2008 to 2017. 2. The severe weather records. 3. Model simulations of the year 2017. Targeting regions: 24 -50 N, 80 -106 W.

Positive and Negative Region Selection storm score: #(event in the bbox) reflectivity score: average

Positive and Negative Region Selection storm score: #(event in the bbox) reflectivity score: average radar reflectivity per pixel Positive regions: 1. Include severe weather events inside. (storm score > 0) 2. For the same event, we label the regions with higher radar reflectivity values. (reflectivity score > thres 0) Negative regions: 1. Far from severe weather events. (storm score = 0) 2. Not having very high radar reflectivity (reflectivity score < thres 1)

Example Red dot: event Green: Positive Blue: Negative Yellow: Neither

Example Red dot: event Green: Positive Blue: Negative Yellow: Neither

Result: Training

Result: Training

Result: Recall-Precision Curve Left: Fix 24 layers in VGG 16 and retrain the next.

Result: Recall-Precision Curve Left: Fix 24 layers in VGG 16 and retrain the next. Right: Fix first 10 layers TITAN Baseline: Recall 0. 43, Precision 0. 21

Application: Severe Weather Prediction with Simulations

Application: Severe Weather Prediction with Simulations

Conclusion We Propose a novel deep learning and computer vision framework to detect the

Conclusion We Propose a novel deep learning and computer vision framework to detect the severe weather events. The design of positive and negative regions for training the network automatically uses the large severe weather label dataset. Future work: validate on the simulation dataset, severe weather classification tasks.

Convolutional Neural Network 20151002 Outline Neural Networks ConvolutionalConvolutional Neural Network 20151002 Outline Neural Networks Convolutional
Neural networks 1 Neural networks Neural networks areNeural networks 1 Neural networks Neural networks are
Neural networks 1 Neural networks Neural networks areNeural networks 1 Neural networks Neural networks are
Neural networks 1 Neural networks Neural networks areNeural networks 1 Neural networks Neural networks are
Deep IDNet deformable deep convolutional neural network forDeep IDNet deformable deep convolutional neural network for
Finger Biometric Neural networks 1 Neural networks NeuralFinger Biometric Neural networks 1 Neural networks Neural
Convolutional Neural Network NVIDIA Jetson TX 1 ConvolutionalConvolutional Neural Network NVIDIA Jetson TX 1 Convolutional
Deep Learning Historical Collections Theory Convolutional Neural NetworksDeep Learning Historical Collections Theory Convolutional Neural Networks
CSCI 5922 Neural Networks and Deep Learning ConvolutionalCSCI 5922 Neural Networks and Deep Learning Convolutional
Moving Object Detection with Deep Convolutional Neural NetworksMoving Object Detection with Deep Convolutional Neural Networks
Convolutional Neural Networks Autoencoders and Deep Learning inConvolutional Neural Networks Autoencoders and Deep Learning in
Deep Convolutional Neural Networks for Image Processing AnDeep Convolutional Neural Networks for Image Processing An
Image Net Classification with Deep Convolutional Neural NetworksImage Net Classification with Deep Convolutional Neural Networks