AI Case Study
Airbus used deep machine vision to detect clouds in satellites decreasing the error rate from 11% to 3%
Airbus used convolutional neural networks to detect clouds in satellite imagery. They decreased the error rate from 11% to 3%, a 72% improvement. They used the TensorFlow ML framework and results were obtained in about a month.
Industry
Industrials
Aerospace And Defence
Project Overview
"...the team at Airbus combined multiple Convolutional Neural Networks and fully connected Neural Network. Similar to VGGnet, by using the GPU as the machine learning engine, which is 40 times faster compared to a CPU, they were able to reduce the training time from 50 hours to 30 minutes. Additionally, by using the HyperTune feature in the machine learning engine, they basically automated all hyper-parameter tuning, which used to be done manually.
The time it took to achieve this was also very short, as the team only spent 1 month before getting promising results."
Reported Results
They reported they reduced the error rate by 72% from 11% to 3%.
Technology
"Multiple Convolutional Neural Networks and fully connected Neural Network". They used TensorFlow as the machine learning framework.
"We showed that the patches with convolutional networks is clearly the best solution for our experimental settings. This choice allows the classifier to learn meaningful features about the spectral content and shape attributes of clouds for classification. A second important result is that the use of superpixels in ANN improves classification performance when compared to pixel detectors."
Function
R And D
Core Research And Development
Background
"...current approaches for cloud detection, that are mostly based on machine learning and hand crafted features, have shown lack of robustness. In other tasks such as image recognition, deep
learning methods have shown outstanding results outperforming
many state-of-the-art methods."
Benefits
Data
Satellie imagery