| |
| A Study On Machine Learning In Muon Tomography |
| |
|
Show general info
| Content: |
Machine learning (ML) has been used in almost all aspects of our lives, but Muon Tomography has been rather slow embracing the full capabilities offered by ML. In this paper, we report on the implementation of ML concepts to depth reconstruction by two neural networks (NN) models. The first NN is the image classification model which is used to classify the tomograms to in- and out-of-focus. The training dataset has around 20 thousand tomograms and is used to obtain the best in-focus image. The second model is the object detection model coupled with k-means clustering for shape extraction. The dataset used in this training is unique and has to be annotated. By using k-means clustering, the shape of the detected objects in the tomograms is extracted to generate the 3D reconstruction. The image classification showed an accuracy of 85% and the object detection model resulted in 80% accuracy. |
| Id: |
3 |
| Place: |
Room: |
| Starting date: |
|
| Duration: |
00' |
| Contribution type: |
Poster |
| Primary Authors: |
BINU, Clive (Texas Tech University)
CHISELENCO, Gabriel (Texas Tech University)
SHANTO, Sadman Ahmed (Texas Tech university) |
| Co-Authors: |
BRADLY, Victor (Texas Tech University)
HOWARD, Madison (Texas Tech University)
WEBB, Katrina (Student/TTU) |
| Presenters: |
BINU, Clive
CHISELENCO, Gabriel |
| Material: |
 Poster |
|
|
|
|
Support
Powered by Indico