Guide: Prof. U.S. Tiwary, Indian Institute of Information Technology, Allahabad
*Collected over 5000 images from 7 different categories. Categories were:
- Homes and Buildings
- 5000 images were classified in 80%, 20% ratio. 80% images were used to train the model and 20% for testing.
- Each category had few subcategories which were trained separately by 7 different models.
- Each layer of ANN was parallelized using Parallel Patterns Library (PPL).
- Artificial Neural Network was coded in C++.
- We achieved 98.6% accuracy.