Skip to content

whoisraibolt/DescNet

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 

Repository files navigation

DescNet

Descriptor Convolutional Neural Network

Tree Structure

src:
| config.zip .json file with all user configurations accessed by training.ipynb file
| inference.ipynb Script to produce some inferences
| training.ipynb Main script to build and training models
| learning_rate_range_test.ipynb Script to test Learning Rate Range
| lr_schedulers.zip Learning Rate Schedulears
|
+---Datasets Folder to stores all datasets used in this project
|
+---Models Folder to stores all models build in this project
|
+---Outputs Folder to stores all outputs/results from models build in this project
|
+---Pretrained Folder to stores checkpoints from models build in this project
|
---utilities.zip Utility Library
|
+---dataset Module to manipulate datasets
|
+---descriptors Module to stores all function related with descriptors manipulation
|
+---global_variables Module to stores all global variables used in this project
|
+---local_feature_detection Module to stores the Local Feature Detection (LFD) Layer
|
+---local_feature_reduction Module to stores the Local Descriptor Convolution (LDC) Layer
|
+---logfile Module to stores all function related with logfile
|
+---reducers Module to stores all function related with reducers manipulation
|
+---residual_neural_network Module to stores ResNet model
|
+---tensorflow_functions Module to stores all function related with TensorFlow
|
+---timer Python Timer Class
|
+---visualizations Module to stores all function related with visualizations and save figures

How to use?

  1. Configure the parameters in config.json
  2. Train the model using python training.ipynb
  3. Evaluate the model on test dataset or custom dataset using: inference.ipynb

TODO:

⬜ Implement Dictionary Learning in scikit-cuda
⬜ Rephrase use of use_desc_block parameter
⬜ Solve MNIST confusion matrix plot problem
⬜ PEP8 and PEP257 Conventions
⬜ Generate requirements.txt file
⬜ Plot ROC-AUC
⬜ Plot Filters
⬜ Plot Feature Maps
⬜ Add RMSE metric to custom objetcs to allow model retraining

Referenes:

About

Descriptor Convolutional Neural Network

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published