Inspiration | What it does
A human being dwelling a life of a deep solitude, a journey of himself constantly striving to find a fulfilment amongst other so alike yet so distant. The discrepancies differ us although aiming for the similarities using the contemporary technologies we are able to find the lacking link through the neural networks. The rest follows up below.
How we built it
We engineered very complicated Neural Network and then simplified it because it required too much memory :P We used LFW Face Database which contains pictures of faces with names.
Challenges we ran into
Finding NN architecture which allows resolving the task. Training good model.
Accomplishments that we're proud of
This project is one of three that we started on this hackathon. We started with predicting number of reactions for posts then we wrote NN for predicting names and last but not least we tried style transfer for music but it was too slow.
What we learned
Processing images in python. Use of convolutional layers. Style transfer for music. Use of fastText for reactions prediction.
What's next for NeuName
Better training on bigger dataset with more names. Maybe we will find some useful application for the model.
Built With
- flask
- javascript
- keras
- lfw-face-database
- neural-networks
- numpy
- pandas
- pillow
- python
- react-native
- tensorflow
Log in or sign up for Devpost to join the conversation.