Class project for Neural network class. I classified the following classes:
- Big Ben
- Boris bus (the red bus in london)
- Shard
- Tube sign
I collected my learning set via google images, I used a chorme extension called bulk image downloader, which made it a lot faster.
I was abel to achive pretty good results in a short time, training took 1 min 12 sec and the accuracy is around 97%
Weights can be downloaded from here: https://www.dropbox.com/s/y17251zl41i1ae0/assignment2_weights.pt?dl=0
I followed the tutorial found on pytorch website and also I found another great tutorial which helped me a lot ( http://pytorch.org/tutorials/beginner/transfer_learning_tutorial.html, https://towardsdatascience.com/transfer-learning-using-pytorch-part-2-9c5b18e15551 )