View Code? Open in Web Editor
NEW
Colab files for running PhD work
colab-gists's Introduction
Within this repo are the 6 different colab notebooks used for Monument Monitor experimental AI work
- 1s First model made with tensorflow. Off the shelf model using semantic segmentation and ADE20k dataset
- 2s Second segmentation model - uses transfer learning
- 3s Third segmentation model - uses transfer learning with different configuration to model two
- 1c First classification model for water levels, using Alexnet
- 2c Second (better) classification model for water levels using pre-trained GoogleNet
- 3c This classification model, that looks creates classifier for all MM case study sites using images submitted up until 1st Feb 2021.
| Model Number |
1s |
2s |
3s |
1c |
2c |
3c |
| Architecture |
DeepLabv3+ |
ResNet101 |
ResNet101 |
AlexNet |
GoogLeNet |
ResNet18 |
| Analysis |
type |
Segmentation |
Segmentation |
Segmentation |
Classification |
Classification |
| Focus |
Area of water pooling |
Area of water pooling |
Area of water pooling |
Level of water pooling |
Level of water pooling |
Identification of site |
| Validation and learning loss tracked |
False |
False |
True |
True |
True |
True |
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent