Evaluating Machine Learning Models for Semantic Segmentation over Cloud Images for Classification

Author

Harsh NagarkarFollow

Date of Award

Spring 4-30-2020

Document Type

Undergraduate Thesis

Department

Computer and Information Science

First Advisor

Yixin Chen

Second Advisor

Dawn Wilkins

Third Advisor

Feng Wang

Relational Format

Dissertation/Thesis

Abstract

Due to the increasing number of available approaches nowadays, choosing the most accurate image semantic segmentation model has become hard. The purpose of this research is to find the best-performing image semantic segmentation model for Cloud classification. For the purpose of this study, a data set of cloud images from the Max Planck Institute for meteorology is used. These images were taken from the by two NASA space satellite.Three main models UNet, PSPNet and FPN were used in combination of 4 differ-ent encoder Inception-ResNet-v2, MobileNet-v2, ResNet-34, and ResNet 101. After training all the models in the Mississippi Center for Super Computing, the results were plotted. Over-all the models turned out broadly similar to each other. Even so, the FPN model with the MobileNet-v2 encoder backbone stood out first followed by the UNet model with the Inception-ResNet-v2 encoder backbone in second place.

Comments

The Evaluation models consisted of three base models

1. UNet
2. PSPNet
3. FPN

Along with 4 Different Encoders

1. ResNet 34
2. ResNet 101
3. MobileNetv2
4. InceptionResNetv2

Crowd-Sourced DataSet consisting of clouds compiled by Max Planck Institute of Meteorology was used.

Recommended Citation

Nagarkar, Harsh, "Evaluating Machine Learning Models for Semantic Segmentation over Cloud Images for Classification" (2020). Honors Theses. 1538.
https://egrove.olemiss.edu/hon_thesis/1538

Accessibility Status

Searchable text

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 International License.