https://github.com/kmohamedalie/dtm_deep_learning_euclid_dataset

Euclid Dataset
https://github.com/kmohamedalie/dtm_deep_learning_euclid_dataset

computer-vision convolutional-neural-networks deep-neural-networks geometry machine-learning mathematics

Last synced: 7 days ago
JSON representation

Euclid Dataset

• Host: GitHub
• URL: https://github.com/kmohamedalie/dtm_deep_learning_euclid_dataset
• Owner: Kmohamedalie
• License: mit
• Created: 2023-04-26T05:03:08.000Z (over 1 year ago)
• Default Branch: master
• Last Pushed: 2023-08-20T10:04:39.000Z (about 1 year ago)
• Last Synced: 2023-08-20T11:26:38.738Z (about 1 year ago)
• Topics: computer-vision, convolutional-neural-networks, deep-neural-networks, geometry, machine-learning, mathematics
• Language: Jupyter Notebook
• Homepage: https://www.unibo.it/it/didattica/insegnamenti/insegnamento/2022/466769
• Size: 26.3 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# **Pattern recognition using CNN**

The Euclid dataset was created by Prof. Guido Borghi at the University of Bologna for pedagogical purpose, for trainig both computer and non-computer science students Machine and Deep Learning in the Digital Transformation Management degree. The entire dataset consist 8000 black&white images with size 224x224 and four(4) different category of geometrical shapes: rectangle, rhombus , square and triangle respectively.

**Complete Jupyter Notebook:** [Link](https://github.com/Kmohamedalie/DTM_Deep_learning_Euclid_dataset/blob/master/Euclid_Dataset_DL.ipynb)









Final output:



Validation accurary: 97.90% 


Training time: 7m  16s 


Resources: Google Colab Standard GPU 


Model Architecture: 2Cov, 2MaxPool, Flatten, Dense, Final output 





![image](https://user-images.githubusercontent.com/63104472/234478794-67bf1708-89d8-4d53-b143-6afc3d815e08.png)




Try the model yourself:

You can download the already trained model load it to your IDE and make prediction: